Bas C. van Fraassen

Princeton University

(this page to be deleted)

2. The three main differences between the Images

4. Deconstructing the Scientific Image

5. Philosophical choices in response

6.1 What is this thing called the Manifest Image?

6.2 And what of that thing called the Scientific Image?

7. The very idea of images

PART FOUR. REAL LIFE WITH SCIENCE

8. A new beginning

9. The continuity of common sense and science in method

10. Perspectival discourse and relativity

11. Value- and function-laden discourse

12. Theory-laden discourse

13. The spirit of gravity vs. the unbearable lightness

Bas C. van Fraassen

Let me begin with a question: how well does science

represent the world? How well does it describe nature, us, and

our relation to nature? Does it give an adequate, exact,

accurate picture, which shows what there is in the world and what

it is like?

This question has a presupposition. It assumes that science

represents, that it gives us a picture, so to speak: the

scientific world picture. This is not an unusual assumption or

way of speaking. Philosophers and scientists themselves have

been writing about the scientific world picture at least since

Galileo, who said that it was a picture drawn by means of

geometry.(2) You may well have recognized this way of talking

from various 20th century writers as well; perhaps you thought of

Thomas Kuhn, Paul Feyerabend, or Paul Churchland, or even

Frithhof Capra. In fact, this way of talking, in terms of world

pictures or world views, comes very easily to us, it seems, it

feels very plausible and natural to speak this way about our

intellectual history.

But that very ease should make us suspicious! If is comes

that easily, isn't it too easy and too good to be true? What

horrors of the intellectual deep are we letting in, as we speak

of this so blithely? What illusions will prey on us, what

muddles are we getting ourselves into?

PART ONE. WORLDVIEWS IN COLLISION (?)

1. The Clash

The question, as I said, has a presupposition, namely that

science represents, that it gives us a picture, or perhaps a lot

of pictures that somehow combine into one: the scientific world

picture. Such a presupposition engenders further questions that

automatically come along with it. Has this picture changed

radically, so that there were perhaps 'ancient', 'medieval', and

'classical' world pictures, while now we have yet a different

one, the 'modern' or even 'post-modern'? Yes, it seems so. In

fact there seem to have been scientific revolutions which

replaced an old picture with a very different new one. Could

there be at any given time more than one rival scientific world

pictures, competing for hegemony? Again, yes, it seems so -- in

fact they could be so radically different as to be

incommensurable. Well, what about pictures besides science or

outside science or before science, is there also a picture we

already have or had about the world, that lives in common sense

so to speak -- the picture of the world as it ordinarily appears

to us -- which still exists side by side with science, but may

eventually be replaced by science in its entirety?

Such are the questions which are brought along automatically

by the presupposition of our original question. If the

presupposition were seen to be false, all its engendered

questions would of course evaporate. To discuss them -- and

eventually that presupposition itself -- I will focus on one

specific philosopher who made all this very explicit. Wilfrid

Sellars presented us with a clear dichotomy: the world as

described by science. which he called the Scientific Image, and

the world as it appears to us, the Manifest Image. Not that the

dichotomy was so novel: Sir Arthur Eddington' famous example of

the two tables is an obvious precedent. The table we see is

solid, it is mostly material even if there are some small pores

and little gaps in the wood. The table science describes,

however, is mostly empty space, filled with small electrically

charged particles frantically whirling around in the void. So

the Scientific Image is astonishingly different from how things

appear to us. Yet science is meant to represent the very same

world in which we live -- and there is the rub.

Wilfrid Sellars he argued that the two world pictures are in

irreconcilable conflict, and that the infinitely superior

Scientific Image must eventually displace the Manifest Image

altogether.

Now I'm going to ask: is this right? What about these

arguments for superiority? What about this irreconcilability?

Is there a real dichotomy, or is that dichotomy itself just an

illusion -- a snare and a delusion created by the smooth talk

that comes so easily to us? And if so, could we not find a

better way to see these apparent clashes between science and the

appearances? Obviusly I am not sympathetic to this 'world view'

discourse, even if I must admit that I fall as easily into it as

the next philosopher, when not on my qui vive. I am going to ask

you to think about rejecting this sort of discourse altogether --

to think about life without a worldview, life without world

pictures ....

2. The three main differences between the Images

The first main difference between the Manifest and the

Scientific Image lies in their history: each image has a history,

and while the main outlines of what I shall call the Manifest

Image took shape in the mists of pre-history, the scientific

image, promissory notes apart, has taken shape before our very

eyes.(3)

The second difference lies in their encoding. In the case

of science we can find a concrete representation: written texts

setting out theories which, even if they have no author, have

many contributors. Are there, similarly, concretely available

descriptions setting forth the Manifest Image? Yes,

Sellars replies: certain philosophers have been writing them. He

refers here to the Aristotelian tradition, which tried to

systematize common sense into a systematic scheme of categories,

but also in our century to the Continental phenomenologists and

in the Anglo-Saxon world to the so-called 'ordinary language'

analytic philosophers. Clearly not all philosophers are engaged

in making the Manifest Image explicit. Some (call them

'metaphysicians') are engaged in quasi-scientific system building

of their own, either continuous with or rivalling both Scientific

and Manifest Image.(4) Other philosophers there are who oppose

systematizing of any sort, engaged instead as intellectual

gadflies or midwives, or intent on showing the flies the way out

of the fly bottle, as Wittgenstein said. Let us therefore give a

special name to those philosophers putatively engaged in

systematic exposition and elaboration of the Manifest Image: call

them the 'systematizers'.

So here are the first two important differences: the

Scientific Image is being created, by scientific theorizing; the

Manifest Image "took shape" in the mists of pre-history, but is

systematically described by the 'systematizing' philosophers.

There is a third difference, which comes to light when Sellars

argues for the former's superiority.

In this enterprise we should, I think, see Sellars' aim

as continuous with Idealism.(5) For to argue the inferiority and

indeed discardability of the Manifest, that comes pretty close to

saying that all we see around us, at least in the way we see it,

is sheer illusion, 'mere appearance' and not reality. This is

not a new theme in philosophy. The British Idealists of the

nineteenth century, classified all we see around us, all we feel

within, the very bodies we have and thoughts we think, as

Appearance. In fact, they mounted various arguments to the

effect that the world as we experience it cannot be real, must be

mere appearance. These are arguments to show that this realm is

full of contradictions -- pursuing our understanding of it we

inevitably find ourselves embroiled in self-contradiction. (Of

course we land in inconsistencies! We are enmeshed in illusions,

in Maya, so what do you expect?) Such were McTaggart's arguments

concerning time, and Bradley's about relations, and many another

wonderful dialectical deconstruction.

Sellars had worked through these arguments and found

them wanting. The Manifest Image -- his version of Appearance --

is consistent, he thinks; but it has other defects. His account

will "compare [the Manifest Image] unfavorably with a more

intelligible account of what there is" (ibid., p. 29). This

sounds modest. In actuality, Sellars attempts more. He tries to

show that the Manifest Image is necessarily incomplete with

respect to explanation -- that it must admit fissures, ruptures,

discontinuities which of their very nature admit no explanation

within the terms of the image itself.

Here emerges, in Sellars' essay, a crucial third

characterization of the two images. The Manifest Image is the

world of a theory which took shape in the mists of prehistory and

which was interiorized by us who (speaking generally, and not

entirely literally) created that theory. But this interior

theory is different from current science not only in its age, but

in that its formation involved no postulation of non-manifest

entities of any sort. The postulational technique of theorizing

is entirely foreign to it.

This is the basis of Sellars' argument that the Manifest

Image will necessarily remain in the position of admitting

phenomena which cannot be explained within it. For sometimes

explanation is possible only by postulating realities behind the

phenomenal scenes. To put it bluntly then, the Manifest Image

must be regarded as Appearance only, and not as Reality, because

it is necessarily explanatorily incom plete. If philosophy has

largely been an effort to systematize the Manifest Image, and is

equally in the grip of the eternal "why?" question, then we

certainly have a clue here to its continual self- destruction.

The 'systematizing' philosopher, if this is correct, tries to

complete the Manifest Image by supplying the explanations it

cries out for, but finds every avenue blocked: any explanation

would involve postulating something real beyond or different from

anything found in the Manifest Image.

This is his first argument, and I will not stop long to

examine it. I have no sympathy with its implicit uncompromising

demand for explanation. Why should we not admit that perhaps

every candidate explanation is a fiction, that perhaps reality

harbors no reasons at all for those phenomena that puzzle us so,

that perhaps the mysteries, as well as the humdrum facts, are

brute?

But I can't leave the issue with this dismissal of Sellars'

first argument, for he has a second argument, to show that the

Manifest Image cannot be of something real. The incompleteness

to which he points is not simply that manifest phenomena lack

manifest causes. Rather, the manifest physical phenomena are

incomplete in the way images and other mental things -- Locke's

general triangle, which is neither right-angled nor obtuse nor

acute, for example -- are incomplete. To this we now turn.

PART TWO. THE PLAGUE OF IRREMEDIABLE VAGUENESS

3. Deconstructing the Manifest Image

Sellars had a favorite example: the pink ice-cube, made by

freezing a soft drink.(6) Within the Manifest Image it is

described as pink all the way through. Suppose you cut it into

finer and finer pieces -- eventually you have pink crushed ice.

But if the very small pieces are separated they look individually

white or colourless -- so perhaps we have to say the ice-cube was

not pink all the way through after all?

Well, trying to elaborate the Manifest Image here, we have

several choices, and different philosophers have tried out all of

them. Placed in a heap, this crushed ice is pink -- so one

option is to say that perhaps the pieces are pink collectively

but not individually? There is another option: the pieces did

not exist in actuality while we still had the ice-cube. The cube

was divisible but only potentially divided, so the pieces only

existed potentially. Hence we could maintain that the ice-cube

was actually pink through and through, though potentially white

or colourless.

In either of these cases we have a problem with vagueness.

For where is the lower bound? At what precise point do we get

collective colour -- or, alternatively, at what small size would

the colour disappear if we perform the division? The Manifest

Image is not given to this level of precision: we can ask the

question, but we won't get a precise answer -- precision would

have to be postulated, and that we can't do here.

Let us be quite clear on this. Whether we think that the

manifest pink ice cube is a continuous expanse of pinkness

through and through, or that it is a vague object whose lower

fineness bound to pinkness is ill-defined, there is no such

object to be found in the Scientific Image. First of all atoms

and subatomic particles are not pink; and secondly, there is

nothing vague, everything is precisely quantified -- if classical

boundaries disappear they are replaced with equally numerically

precise probabilities, and if those disappear they are replaced

by exact sets of probability measures, and so forth. The two

images are of worlds which cannot both be real, for as described

the pink ice cube cannot be identical with any object in the

world described by science.

What Sellars is denying here is that the Manifest Image can

be accommodated by science, that it can be reduced to something

scientif ically respectable. It can be replaced, but it cannot

be recon structed or reduced to something in the Scientific

Image, for any reconstruction or reduction would distort or

change or improve, it just couldn't leave it the same. However

we try to explain the way things appear to us, we run up against

the openness of ordinary language. The assertions we make in our

ordinary language is full of vague promises which we know we

cannot make good on -- but life is like that.

When the openness is irremediable, within our own terms,

does it not follow that we literally don't know what we are

saying? Metaphysics and science, on the other hand, with their

regimented languages, precise concepts, and quantifiable

distinctions, appear to provide new terms in which the openness

is remediable. ... a framework where vagueness or unstated

qualifications are at most a practical defect, in principle

removable. There we can speak responsibly, by the strictest

standards, for the first time. Or so, at least we may hope ....

4. Deconstructing the Scientific Image

But now, with that problem in mind, let us take a close look

at the Scientific Image. The revolution of Renaissance science

and its codification in the seventeenth century aimed to remove

these defects from our world picture once and for all. The

primary qualities are really quantities, exhaustively described

with full numerical precision in analytic geometry and

differential calculus.(7)

But science has higher standards of precision, and so, when

it comes to discussing vagueness and indeterminacy we have to

hold the Scientific Image to much higher standards than the

Manifest. Those higher standards are proper to its examination

exactly because it set itself so much higher standards, namely

those of mathematics. We should raise questions concerning the

Scientific Image proper to it, of a sort it would have been

unfair to raise for pink ice-cubes: mathematical questions.

Consider this beer glass: it has a shape. What that shape

is, precisely, we do not know. It was assumed that it is an

analytic function of the spatial coordinates (in the way that a

straight line "is" a function y = ax + b). It has one shape, and

that shape is a geometric object; with equal justice it is a

function defined on the continuum of real number coordinates.

We are speaking here of the continuum of classical

mathematics which has equal use for the representation of each

primary quality: length, duration, shape, size, number, mass,

velocity, what have you. The equation of the primary quality

shape with geometric shape -- on which Galileo placed such

emphasis -- is in effect the assertion that a certain

representation is completely adequate. But now we must ask:

what exactly is this representation?

Well, shortly after Galileo, Descartes created analytic

geometry, in which shape is represented in the way I just

explained. But you have to realize that what he created was not

exactly the analytic geometry we have today. For example,

Descartes allowed only finitary constructions in geometry, so a

point only exists if two lines are constructed to intersect

there. It was his contemporary Pascal who, very controversially,

insisted on the ubiquity of the infinite, and said that a line or

a plane is composed of infinitely many points. So the beer

glass' shape already had rival representations at this early

point. In the nineteenth century mathematics had developed much

further, and it was sensible to ask: is this shape an analytic

function? There is no question but that, as a reconstruction of

the world picture of Galileo, Descartes, and Newton, we can

choose either option. They had not said that every physical

magnitude in nature is an analytic function, but they had not

conceived of any alternative. Nothing would have been lost from

the subject as developed so far if we had added to it that all

the functions describing the primary qualities of real physical

things are analytic -- nor if we had added that some are not.

The description was open, indeterminate in that respect. Nor was

there any kind of experimental evidence to cite. The only

questions asked are, it seems, about which options could lead to

more fruitful developments in later physics.

If we go on to still later mathematics, the strange and

previ ously unaskable questions multiply. Around the turn of the

century, Lebesgue and others developed measure theory. This made

it possible for Birkhoff and von Neumann to raise a new and

interesting question about the shape of the beer glass.(8) They

pointed out that when classical mechanics solves problems about

systems with given precise configurations, we can construe it as

using conveniently simplified descriptions. For those

descriptions will distinguish between regions that differ only by

point sets of measure zero -- ones that are not empty but

literally have no lenght, no area, and no volume. More

realistic, they suggested, would be the description that results

if we transform the precise descriptions by identifying regions

that differ only by sets of measure zero. Their reasons for

thinking of that as more realistic may or may not be cogent, but

it suffices here to note the conceptual possibility. That is,

after Lebesgue we can look back to the older description of

nature and we have a new option. We can accept or reject the

following advice: "Let the calculations go on as usual, but the

shape is correctly represented not by one region in geometric

3-space, but by an object in the quotient construction that

identifies regions modulo differences of measure zero".

You will realize that I am simply giving examples of how, in

many ways, we must in retrospect look upon the Scientific Image

inherited from the older generation as open, vague, ambiguous in

the light of our new understanding (that is: in the light of

alternatives not previously conceived).

What is the shape of this beer glass really? What was it in

the Galilean, Cartesian, Newtonian Scientific Image? Indeed, we

need to cast our net more widely still, if we want to find all

the ways in which we could now understand the Scientific Image

fashioned in the seventeenth century. There is no such thing as

the classical continuum, if that is meant to be the continuum on

which the classical (= modern) Scientific Image was erected

originally. Cantor, Brouwer, and Weyl had equal right to regard

it as erected on their continua, which are very different. Of

course, today we will use "the classical continuum" to refer to

the subject of real number theory as it now exists in main stream

mathematics. That is the politics of linguistic usage. But

their are these alternatives, which can within what we now call

classical mathematics be regarded as perfectly well defined

mathematical objects.

So, what would you like the shape of the beer glass to be?

The openness of scientific description here come to light is

irremediable. Of course, every time we outline a range of

alterna tives for ourselves, we can ascend our private throne --

are we not all kings and pontiffs in realms of the mind? -- and

assert that one of these alternatives is the one true story of

the world. When the range of alternatives is refined by new

conceptual developments -- or simply by having our attention

drawn upward by logical reflection -- we can choose a new option

and make yet another declaration ex cathedra. Arbitrary perhaps,

but as definite as can be, by choice. What we cannot pretend is

to be non-arbitrary, or to close our text once and for all.

Yet the form of understanding is always one of presumed

objectiv ity and univocity. The Scientific Image is as replete

with uncashed and ultimately uncashable promissory notes as the

Manifest Image.

5. Philosophical choices in response

We have gone some way now to dispelling the air of

superiority of the Scientific Image over the Manifest; but

nothing I have said need necessarily be seen as a disaster for

either. With respect to vagueness and ambiguity there is at most

a difference of degree; there is no difference in principle, and

if there is a problem of principle then the two images are in the

same boat. What does follow is that anyone taking seriously

either our ordinary way of understanding the world, or the way of

science must take vagueness and ambiguity very seriously. The

lesson learned in these reflections is that vagueness is

irremediable, in science as well as in pre-theoretic description.

Accordingly, this vagueness is, in itself, no defect, though one

might wish to opt for or privilege the less vague image.

One option would be to insist that one of these images, or

one of their possible successors in the course of human history,

is actually a complete, non-arbitrary, correct representation of

what there is. The world is vague; our task it is to develop

conceptual tools adapted to the non-distorting description of a

vague world.

Another option is to postulate that at each juncture, one

refinement that diminishes vagueness will be more accurate than

its rivals. The world is sharp, but impossible to represent

sharply; that sharp world lies jenseits aller Vorstellung, beyond

the endless task of constructing its image(s).

Both options, however, commit us to leaving the basic

dichotomy of images intact. If we adopt either, we would be

deeply engaged in metaphysics, and find ourselves on one side of

a very deep divide in that putative enterprise. But can we

really be so complacent about this dichotomy? I say not: the

difficulties which this account of incommensurable worldviews led

us to, seem to me to indicate something much more fundamentally

wrong with the entire dialectic. That any description will

always, upon a little pressing, turn out to be vague and often

ambiguous, that every text is open, that despite all

philosophical ambition no one can produce a text invulnerable to

deconstruction -- that, I think is definitely so. But this

philosoph ical story of images and worlds, perspectives,

conceptual frames, and all their ilk is not thereby shown to have

a coherent fall-back position in a metaphysics of vagueness. The

flaw in Sellars', or any, story of clashing worldviews may lie

much deeper.

PART THREE. AN INCOHERENT FICTION

When all the answers available lead you into absurdity, Kant

argued in a famous passage, it's time to examine the

presuppositions of your question. For a question may itself have

something wrong with it, and thus make all its own answers

impossible.

In our present context, it would mean this: there are no

such things as the Manifest and the Scientific Image at all. Is

that possible? Yes, in fact I can think of some very good

reasons for that conclusion. If you agree to them, you may even

find some reason to generalize this sceptical conclusion to all

those -- what shall I call them? -- world-pictures, conceptual

frames, worlds (as in "the world of science", "the world of the

physicist", "the Ptolemaic world") which have so easily and

smoothly crept into our discourse.

6. The Images as philosophical miscreants

My first two reasons concern how Sellars has misrepresented

both our ordinary understanding, when we are not consciously or

even implicitly drawing on science, and also science itself.

6.1 What is this thing called the Manifest Image?

The Manifest Image is the way the world appears to us; it is

also the world as described by the 'systematizing' perennial

philosophy, and it is the image to whose evolution and

development all postulation remains forever foreign. There is

nothing that fits this description. The philosophy in question

engages freely in reification and postulation of all sorts.(9)

Putative entities like conceptions, conceptual frames, images,

and worldviews are indeed introduced within the perennial

philosophy, through the reification of the language forms we

create in such easy profusion -- but that is exactly what

disqualifies the perennial philosophy from describing something

to which postula tion is foreign.(10)

Should we say then that those philosophical

descriptions are simply faulty accounts of the Manifest Image as

it really is? Unfortunately we have no description at all of

that Image except by the philosophers Sellars singled out as

engaged in that enterprise. Can we take Sellars' own initial

description -- the Manifest Image is the world as it appears to

us -- as the definitive identification? Could we in fact say

that in this phrase, the Manifest Image is introduced into the

philosophical pantechnicon by explicit definition?

Now here we encounter the philosophical "as". This

"as" is really the same as the infamous philosophical qua, a

hyper-intensional locution of dubious intelligibility. A

description of a thing may be correct or incorrect -- what is

denoted by "the thing as described"? Something that exists

regardless of whether the description is correct or incorrect?

Or does it simply denote the thing, if correctly described, and

denote nothing at all otherwise? On the latter option, if

"Manifest Image" and "Scientific Image" are not denotationless,

they denote the very same thing, thus ending all philosophische

Spitzfindigkeit at once. But with the former option we would

commit ourselves to an ontology which most of us -- including

Sellars -- would explicitly reject, and for which he takes no

responsibility.

So one side of this dichotomy is simply a self-created

muddle, designed to give us a house of cards ready to fall apart

under the scientific stare.

6.2 And what of that thing called the Scientific Image?

What about the Scientific Image? Isn't that at least real,

and don't we have to confront it, cope with it, and relate

ourselves to it every day? And does it not, by its very design,

omit those colours, textures, smells, feelings and emotions,

drives and aspirations that constitute our human existence?

Isn't that reduction to the physical minimum our heritage from

Galileo's insistence that science proceed entirely in terms of

the quantifiable 'primary quantities, which set the program, in

essence, for all future science, all the way to our day and for

what we still expect in our future?

Once again, what I see here is something designed with the

resources of rhetoric, that bears little relation to the actual

history to which it appeals.

When Galileo insisted that science restricted its

descriptions to a very few primary qualities, he had a good

point. One of the defects that rendered the Scholastics'

scientific tradition less and less effective was the the

unconstrained multiplication of properties which passed for

theorizing among them. So this innovation of Galileo's was much

needed discipline. Compare this practical point with the later

philosophers' reading of it: as a move introducing the great

divide, the separation of those properties which do really

pertain to the systems described by science -- the Scientific

Image -- and all those other properties of our acquaintance which

do not belong there. Scientific discipline did not require that

idea!

Galileo himself was to blame. He could simply have claimed

certain theories to be true and left it at that. Galileo was

not so modest. A complete description of nature would give all

its quali ties, both primary and secondary -- but the latter, he

claimed, could all be reduced to the former, so that science

[the theory framed in terms of primary qualities only] would be

complete. In this conten tion he made two dubious moves, neither

of them vindicated by our later history. Firstly, there is his

completeness claim for the total list of properties -- which

all, at that time, were humanly sensible properties, very

different from what science eventually marshalled as its basic

theoretical quantities. Secondly, there is his claim of

reduction. In fact, very little of the generally accepted

description of the world at the time could have been given simply

in terms of that list of properties; nor could it be now!

6.3 The dialectic that engenders the dichotomy

At this point you can see the dialectic moving with its own

inner necessity. If B is not reducible to A, then either A is

incomplete or the two are incompatible. So if A purports to be

complete, then either it is false or else B is false -- one of

the two must be eliminated. Here we have the picture: there are

two putatively complete images of the world, and they are

incompatible.

But remember how I introduced this dialectic: Galileo's

restriction of science to the primary qualities was a very good

practical move for science, because it imposed a much needed

disci pline on scientific theorizing. What does that highly

practical and commonsensical endorsement have to do with the

ensuing dialectic?

This dialectic can persist only through the maintenance of

an illusion. That is the illusion that "the scientific

description of the world" or "the primary qualities" refers, and

keeps referring, to one definite subject. Look at Galileo's

primary qualities. He was still a bit soft; Descartes was the

master of discipline, and made the cut at the only natural joint

in sight, namely, the quantities definable in terms of spatial

and temporal extension. But what happened when these were

demonstrably not enough? Scientists understood the idea of

discipline better than philosophers: at that point they very

common-sensically introduced additional primary quantities. In

the centuries that followed, not only did they repeat this

manoeuvre as needed, but they also changed the original list,

replacing spatial and temporal quantities by spatio-temporal

quanti ties for example.

So the exclusion from the scientific vocabulary is a

practical matter, it is provisional exclusion, not a matter of

ontological principle. We abstract as far as we can to strip our

problems to the bone, so as to see through the superfluous flesh

to the skeleton -- but when we encounter new problems, we may

have to retrench a little on that scorched flesh policy. That is

not only the practical way to proceed in science; science must

have learned it from practical men and women. When something is

provisionally excluded, that is with the idea that eventually

either (a) it will be shown to be reducible to what is included,

or (b) it will itself be introduced into the scientific

vocabulary, or (c) we will find that something new is introduced

to which it is reducible.(11)

The argument for scientific realism from the

incompatibility of Scientific and Manifest Image -- given the

imperative to maintain the correctness of the science we have

accepted -- is therefore disingenu ous. For from the point of

view of science there is no incompatibil ity, there are only

temporary sticking points. Adjustments will be made on both

sides, as need be, so as to reach accommodation. There are no

stable A and B which have proved to be mutually incompatible.

The argument is disingenuous in another way. For the

complete ness claim which is crucial to the argument for

incompatibility is itself a dialectical miscreant. First of all

(this is related to the preceding point) it is infinitely

malleable in content: no one claims completeness for current

science, but only for science in principle in the ideal long run.

Since no one can know what that will be like, no one can know

what is being claimed in this completeness claim. But secondly

(this is a new point) the completeness claim does not come from

science, it is the philosopher's distorted codification of

certain laudable aspirations in science. The scientific project

is to reach a point (as Nancy Cartwright puts it(12)) of

predictive closure. Descartes thought that he could develop a

deterministic pure kinemat ics; but the true kinematic

descriptions at t+d cannot be predicted from the true kinematic

descriptions at preceding time t. Therefore the list of

quantities is increased by Newton, to include dynamic quantities;

and it seems that closure is attained. But predictive closure

does not imply descriptive completeness -- that was not even the

aim!

7. The very idea of images

Very striking in Sellar's characterization of the manifest

and Scientific Image are two facts that should have greatly

worried him.

The first is this: Sellars had said that he would explain

his terms, but was then content to do so in the language of folk

psychol ogy. That is the account of human nature which

introduces such mental entities as images and conceptions that

populate the world together with platforms and Constitutions. So

when he explains what sorts of things these images are, he

resorts to terms belonging to traditional philosophical

psychology and to folk psychology -- all stuff that finds no

place in the Scientific Image, unless it be the place of

phlogiston, N-rays, entelechies, and cold fusion. Note well: I

am not disparaging psychological discourse here; I am saying it

is not reducible to the discourse of physics, and Sellars cannot

help himself to it in this context.

The second fact to be noted is that by his own account,

within the Manifest Image introduction of such ideas as these --

that there are these images, world-pictures, conceptual frames or

what have you - - counts as postulational and is therefore by

definition foreign to it! In telling his story of those images,

Sellars was therefore speaking from a perspective located neither

in the Manifest Image nor in the Scientific Image -- thus,

according to that very story, located nowhere at all. So Sellars

is, as it were, speaking from within an ontology which he does

not make explicit, which in effect he had already implicitly

disowned, and for which he takes no responsibility.

Finally, let us be quite blunt, and bear down on this term

"image" itself. We know very well what an image or picture is,

in the primary usage of that term -- we see such things every

day. But here the terms are of course used analogically. The

effect of the analogy was to suggest that the philosopher is not

thinking about real things but about a humanly created "likeness"

(picture, graven image, description) or alternatively some

naturally arising "likeness" (after-image, reflection in a pool,

fata morgana).

To draw an analogy is only to say that it is "as if", and

that we may gain some understanding from focusing on one respect

in which two things are alike. But this particular analogy is

apparently used to reify, to introduce an entity [indeed, two

entities] which are like pictures. What sort of entities are

they?

Perhaps you would like to say that these images must be

things existing in the mind, mental images, mental entities. I

do not know how far you are willing to trust this sort of talk,

whether as part of folk psychology or in some more technical

guise within cognitive psychology. But we have for a long time,

at least since Wittgenstein, found it impossible to rely on it

uncritically. You may know Wittgenstein's demonstration that the

very idea of a mental image makes it something fundamentally

unlike a real image, so that the analogy pretty well destroys

itself. This is his demonstration from the so- called

"duck-rabbit" picture, an optical illusion which is seen

alternately (and quite spontaneously) either as a duck or as a

rabbit. This sort of phenomenon is what supported the idea of

mental images, for the explanation offered was that when two

people look at the real picture, and see something different,

they have different mental images. For this sort of explanation

to work at all, we have to say that a real image is something

that can be seen in two different ways, while a mental image is

something that can only be seen in one way. But it is crucial to

the very idea of a real image that it is something that can be

seen in different ways -- so, conclusion, mental images aren't

images at all.

PART FOUR. REAL LIFE WITH SCIENCE

Perhaps you accept this, and say fine, Sellars told a little

fable to draw attention to something important. Images,

conceptions, categorical frameworks, world-pictures are

themselves fictions that facilitate the discussion of something

really important. Since this very way of talking, if taken so

literally, seems to lead us into incoherence, let's not take it

too seriously, but concentrate on what is important. Important

is the crucial insight: the insight into the impossibility of

reconciling science and our ordinary common sensical way of

thinking. Well, if you are so compliant, let us see what follows

from this. We have to start all over again! What does the clash

of images, their vagueness, and so forth, amount to if there are

no images?

8. A new beginning

Many philosophers separate science sharply from ordinary

life and ways of thinking. With such a sharp separations, our

options reduce to extremes.

One option is instrumentalism, while attempting to

place our ordinary way of thinking on a pedestal and preserve it

through isolation. This ignores the fact that our response to

our experience never takes a necessary form but is a historical

product that could certainly have been different, and is in any

case subject to constant change. The option is at war with

itself for it purports to safeguard our history by abrogating the

historical process.

The other extreme is scientism: if science is radically

different and also superior, then we must submit ourselves to it

wholly, forsaking all others. While the first option ignores the

historical origins of our ordinary way of thinking, this one

ignores the equally checkered historical development of science.

Science has never enjoyed such undisputed superiority, has never

ceased discovering its own shortcomings, and can't pretend to a

faultless process of self- perfection or self-purification.

It seems to me that our verdict for both extremes

should be the same. Not only is reification of worldviews a

highly theoretical move of doubtful internal coherence, it stems

from a radical misconception of the human condition.

If you ask me how things seem to us, I cannot do anything

but speak and write. There is some choice: I can either invite

you to observe the way I speak and write in response to my

experience, or I can describe to you how things seem to me. On

the first choice, you will see and hear me using the language of

daily life -- some of which could of course be life in a

laboratory if I am a scientist. This language certainly does not

embody perfect understanding, you will detect some

misunderstandings and some lack of understanding both in the

language itself and in my use of it. You'll spot the defects all

the more easily if you are aware of theories and myths that have

played a role in our history, for those have certainly been

factors in the evolution of our linguistic practices.

The defects get considerably worse, however, if I choose the

option of describing how things seem to me. If I give you a

philosopher's account, it will be pretty medieval, full of

disposi tions, possibles, potentialities, universals, and the

like. If I give you a scientific account, whether from

psychology, physiology, or physics, you'll notice that the feel

and taste of real experience just is not there. Science is

driven by highly practical motives. For that reason, the

scientific account slashes and burns, to eliminate all factors

that do not contribute to meeting its own criteria of success.

That is only right, and as a practical person I applaud it - -

but then cannot understand the philosopher who insists that the

scientific account must be the one that is complete, that its

sparsity is simply irredundancy with respect to all criteria for

adequate description.

Yet, as with all great philosophical mistakes, there must be

something to it: For every one of us there is therefore some

point of rupture between, as we are inclined to say, the way we

see the world and the way science describes it. On the other

hand, we have the impulse to say with great conviction something

that we can't seem to disentangle from metaphor but insist on

nevertheless: that on a certain, familiar level, we would be in a

position to communicate with all our forebears and descendants,

that we can reach through all cultural differences to the shared

human and earthly reality beneath.

Could we possibly, ever (now, finally?) discuss this without

slipping into metaphor at every turn?

9. The continuity of common sense and science in method

What of Sellars' noble savage who lives, moves, and has his

being in the Manifest Image? We have never been like that. The

great and crucial divide, according to Sellars, is that

scientific world views are fashioned by postulation while the

Manifest Image contains nothing postulated, only things

experienced though misdescribed. Common sense, ordinary

thinking has just one major dynamic principle, and it is

superstition. The tactics and gambits of superstition are

exactly analogy, metaphor, and linguistic extension followed by

personification and reification, thus furnishing the world with

vast arrays of newly postulated entities. Its driving force is

the demand for explanation and the satisfaction derived

therefrom. Inference to the most probable conclusion or to the

best explanation are endemic in the tabloid newspaper, books

about UFOs, the chariots of the gods, the miraculous efficacy of

herbal cures, and so forth.

Of course the description I just gave of the mechanisms of

superstition bears some likeness to various philosophical disqui-

sitions on the structure of science. Nor need those be wrong:

superstition, rational common sense, and science may have much in

common. In fact, I was describing superstition here exactly in

order to argue that Sellars' description of life in the

Scientific Image fits all life, including that of the pre- and

un-scientific -- not in order to convict us of irrationality.

But there is a difference: that in science these processes are

bridled, constrained, checked in their course by harsh demands of

productivity -- which they are much less, and never

systematically, in ordinary life. Science is bridled

superstition, just as rationality is bridled irrationality.

So there is a clash, yes: bridling the unbridled meets with

opposition. Science teaches us how not to believe things, how to

let go of our ideas; but we love and cherish our ideas and their

security. Rightly did Isaac Levi speak, in his epistemology, of

relief from agnosticism. But note well: this bridle is not the

yoke of a foreign prince, imposed in alien fashion from outside.

Rather, if within our common sense we reflect on ourselves, we

already applaud such bridling.

10. Perspectival discourse and relativity

There are many differences between 'ordinary' and scientific

description. The first is that ordinary description is always

perspectival, for obviously practical reasons. But such

perspectival descriptions are banned from theoretical science.

Here we have in a nutshell the idea of relativity: as soon as

tacit relativity is detected it is first made explicit and then

banished in favor of the (more) absolute. (Hence the irony of

lumpen relativisms' air of warrant from science.)

But we must make a crucial distinction here, easily

illustrated by what is perhaps the earliest illustration of such

a theoretical change. The first astronomical frame of reference

is the observer's Zenith and horizon. But already in ancient

times, its relativity was realized. Hence there was a shift to

the North Star and the Celestial Equator as frame of reference,

which is independent of the observer.

Now the distinction: the relativity detected was clearly

not precisely observer-dependence but rather location-dependence.

In order to use the description given in the common, "absolute"

frame of reference, the observer still has to locate himself

therein, so he still needs to use perspectival, or to be more

precise, indexical language: "I am there, here is my Zenith,

this region is within my horizon". This perspectival or relative

form of description cannot disappear from science if it is also

to be applied science. But in theoretical science, there is no

such indexical description, and the location-dependent

description is replaced by location-independent description.

There are two wrong reactions when intellectual reflection

has brought to light a new and still farther reaching relativity.

(We have seen this very clearly illustrated in the two

well-documented philosophical reactions to Galilean and

Einsteinian relativity.) The first is, obviously, denial: "No,

there is absolute simultaneity and length, it is simply not

describable in the language of Einstein's physics". The second

is sickly affirmation, a bee-line for a new security: "Space and

time are unreal, simultaneity and length are characteristic only

of objects-of-thought, of the world we pictured to ourselves

which turns out not to be the real world. Only what is invariant

under the newly understood group of transformations -- Galileo's,

Lorentz's -- is real. We lived in Maya, created by our own

minds. Develop process metaphysics! Abandon persistents,

develop an ontology of time-slices, punctal events, space-time

worms"! I say, do not heed these counsels of despair. The only

authentic reaction is the one that happens quite naturally in

practice: nothing is given up, no form of assertion is discarded

as meaningless, though of course we have now a richer and more

nuanced construal of what we used to say. That is to say, the

very same 'local', 'perspectival' descrip tion is now related to

a different theoretical model.

Just as ordinary thoughts about the pink ice-cube were never

(except in the philosopher's fiction) wedded to

pinkness-through-and- through so ordinary thought was never

wedded to a denial of Einsteinian relativity.

11. Value- and function-laden discourse

There is another source of apparent conflict between science

and experience: our ordinary descriptions are charged with value

and emotion, with needs, intentions, goals, and instrumental

evaluations relative to those goals.

At first sight, ordinary naming and classifying seems

largely use-independent. That may be so; but the dimension of

praxis reaches for deeper than might be at first apparent. What

about, for example, "tree", "rose", "lettuce"? Are these

ordinary nouns completely characterizable without reference to

praxis or intentionality? Well, roses are flowers; you may tell

me that you gave your mother roses, or equally appropriately say

that you gave her flowers. So why not, if you like, just tell me

that you gave her pieces of plants?(13)

This use-related character of discourse is of course

evident also in the laboratory (as is perspectival,

'pre'-relativistic discourse). Things are called by names that

relate to their function, not to their physical constitution,

when scientists work. The disparity with theoretical discourse

is then all the more blatant. Neither in pointing to indexical

language nor when I mention value-, use-, and function-laden

discourse am I contrasting the language of the scientist with the

language of the layperson. Both are indispensable to us, both

inside and outside science, exactly when we turn back to those

principles and constructions we have made as 'objective' and

impersonal as we possibly can, in order to draw on them for

living and acting.

Is there a clash here? Only apparently so. Our ordinary

discourse is not reducible to theoretical descriptions in the

language of physics, even if the latter is complete within its

own terms of reference. Within science as activity, the two

forms of discourse are happily integrated. That activity

includes after all, besides the construction of theories and

models in all their pristine purity, our use and application of

those pure beings in our practice. That theoretical description

does not pay heed to the location and interests of the speaker is

just right. It does not mean that values, use, and

intentionality exist only in some rival to what theoretical

language describes. Nor does it mean that the theoretical

description is factually incomplete; it means that theoretical

language has a limited use. Its resources are not sufficient for

ordinary discourse, not even for applied science; but they are

not meant to be.

12. Theory-laden discourse

Now let me admit to one genuine source of conflict

engendered y scientific theory change. It is true that language

is always theory- infected, loaded with assumptions of all sorts.

Consequently, when a new scientific theory comes along,

contradicting older such theories and also common assumptions, it

pulls the rug from under the way we speak. First it cleanses and

then it infects our language in its turn .... Metaphors aside,

this is surely so, since some new words are brought to birth in

the laboratory, theoretical monograph, and patent office.

Let us, for simplicity, imagine that the radio was patented

by Edison, and that the patent description is in terms of

vibratory wave- like motions in the aether. This is where the

new word "radio" received its meaning. A device is a radio if

and only if it satisfies that description. Today's science says

that the aether does not exist. So, anyone who believes current

(1995) science and claims to have a radio is contradicting

himself -- right?

Well, language is a little more complicated than that.

Language is more like a wily, survival-adaptive animal than like

a machine. The word "radio" left the patent office, forgot its

theoretical origin -- or was adopted by a society happily

oblivious of those theories -- and continued to flourish well

after its original meaning turned out to exclude everything from

its extension. As soon as the word "radio" became common coin,

the criteria for application in common use were relaxed -- and

those relaxed criteria obviously had priority, they alone seemed

to matter when the "defining" theory was given up. Dictionaries

are updated; patent law too is flexible. How the judge would

laugh if lawyers tried to argue infringement of patents on such a

theoretical basis!(14)

I chose this example only partly to show that there is real

conflict here. Such a case as this is in fact a prime example of

how ordinary language can become theory-laden. At the same time,

it shows how needlessly overblown is the dichotomy of Manifest

and Scientific Image. It is true that assumptions and theories

get 'embodied' in our language, that there are theoretical

presuppositions of applicability even for very common nouns. But

this is not a clash between ordinary and scientific thinking. It

is a type of clash to be expected equally within Sellars' and

Churchland's ideal scientific speaker community of the future, as

well as within the most illiterate pre-scientific society.

Behind many puzzles over the clash of appearance and

scientific description lies the conviction that communication is

impossible or seriously hampered if conducted in a language laden

with a false theory, or with a theory not believed to be true.

In the original sense, there are no radios; but no one noticed.

So if one person used "radio" in an attempt to refer to a real

thing, other people, relying on the same false beliefs, took him

to refer to exactly the thing he meant. But furthermore, when

they all realized that there were no radios in that sense -- and

perhaps had as yet no new, accepted theory to replace the

original definition -- they kept using that same word to refer to

real things and kept communicating successfully. The adjustment

was, at least, pro tem, a small bit of semantic ascent. For if

someone said "radio" everyone took him to refer to those things

which were classified as radios under the now rejected theory.

There is therefore no difficulty in principle in simultaneously

saying that you doubt the existence of the aether,

electromagnetic waves, electrons, etc. and describing the objects

around you as radios, VHF receivers, computers, electric lights,

and so forth. A good theory of language must be in accord with

this, and shed some light on this.

13. The spirit of gravity versus the unbearable lightness

What a state of affairs we are in! Doesn't it cry out for

metaphysical labor? At such moments, when the language in use is

laden with doubted theories, discredited old assumptions, and

already given up beliefs, we do not have a coherent opinion at

all. Common sense has become a hodge-podge, laden with

ontologies that fit only long discarded scientific views,

hobbling along on make-shift metaphor and hastily carpentered

crutches. Metaphysicians must set to work and show us how to

cleanse, regiment, and elaborate a new system of beliefs,

together with a language laden at most with the stablest of those

beliefs. We need a coherent ontology, fit for science and

accommodating common sense, a worldview in which we can rest in

peace. Do you agree?

I do not. There is clearly a lot to be said for

straightening out our concepts 'locally' -- for example, those

involved in our beliefs about the pens, pencils, and writing

paper we use every day, the roads we walk, the rocks we climb --

to the extent appropriate to our immediate goals.(15) The

question is: are we in poor condition if we do not do so

'globally'? That is, if we do not achieve unlimited cleansing

of our language -- the entire description of nature and our own

place in it -- from presuppositions that we do not fully believe.

What could be the argument to the effect that, prior to

success in such a far-reaching enterprise, we are in poor

condition? One premise might be that local efforts of the same

sort cannot yield a coherent view if made within an overall

defective context. But that, I think, is false. We live in that

conceptual quicksand -- morass if you like -- we dance on that

sort of tightrope fastened to highly suspect supports, we do

build on sand, and look! we function perfectly well! A second

premise might be that it is possible to succeed in that global

enterprise, and that it is a project worthy of one created only a

little less than the angels. (A work worthy of a man, as one

might have said only a generation or so ago.) But here I beg to

differ. Not only does it seem clear, from the actual structure

of our existence, that we flourish while lacking any coherent

world-view. It seems equally clear that the proposed global

representation of beliefs and cleansing of language is literally

impossible.(16)

From this I draw uncompromisingly the consequence: clear

thinking in local matters does not require that we have, either

actually or potentially a global conceptual scheme, metaphysical

system, or worldview. A task more worthy of philosophy than the

spinning out of such systems is trying to understand how this can

be. That is the task of defeating a Spectre which claims the

consequent utter meaninglessness of all thought. It is the

problem all of us have, being post-foundationalist, post-modern:

to describe ourselves without resorting to or falling into what

Kant called the illusions of Reason.

Where exactly does Aristotle describe walking? If I

remember it rightly, he says that we keep our center of gravity

over one foot while moving the other to a secure place, and then

shift our mass. This would indeed be prudent! But it describes

a sort of goose-step, not our real walking which is a continuous

falling forward, a slow version of a headlong run, trusting

ourselves to fortune. Learning to walk is learning to fall.

--------------------------------------------------

ENDNOTES

(1) This paper was presented earlier as part of the James B.

and Grace J. Nelson Lectures, University of Michigan (Oct. 1994),

and of the Kant Lectures, Stanford University (Apr. 1995) as well

as at the Einstein meets Magritte Conference (Brussels, May

1995). For earlier thoughts on this subject, see my "On the

Radical Incompleteness of the Manifest Image", PSA 1976, vol. 2

(East Lansing, MI: Philosophy of Science Association, 1977), 335-

343 and "Critical Study of Paul Churchland, Scientific Realism

and the Plasticity of Mind", Canadian Journal of Philosopy 11,

(1981), 555-567. I would like to thank Prof. J. van Brakel for

helpful comments and discussion; his "Empiricicism and the

Manifest Image" (ms. 1995) includes a response to my view as well

as to an extensive ambient literature (see further note 9 below).

(2) The recent popularity of such terminology, however,

appears to begin with Hertz in the late nineteenth century.

(3) "Philosophy and the Scientific Image of Man", chapter 1

in W. Sellars, Science, Perception, and Reality (New York:

Humanities Press, 1963), p.5.

(4) The reader may suspect that there is not such a great

difference between the two classes which I'm calling the

'systematizers' and the 'metaphysicians'. My nomenclature tries

to follow Sellars' typology here, and we'll have to see whether

it is well based.

(5) See "Philosophy and the Scientific Image of Man",

section V, p. 26ff (especially p.29).

(6) See "Philosophy and the Scientific Image of Man", p. 26,

which is however just one of the places where this example

appears. See also for example the section "A Pink Ice Cube" in

Lecture 2 of Pedro Amaral, The Metaphysics of Epistemology:

Lectures by Wilfrid Sellars (Atascadero, CA: Ridgeview Pub. Co.,

1989), and section V of "Scien tific Realism or Irenic

Instrumentalism", R. S. Cohen and M. Wartowsky (eds.) Boston

Studies in the Philosophy of Science, vol. II (New York: 1965).

(7) There was just one voice in the wilderness: Berkeley

arguing that the primary qualities were not originally any better

off than the secondary ones. I do not want to examine his

argument here, but I will state in contemporary terms what I take

to be his conclusion: the privileging of primary qualities and

their geometric representa tion was an act akin to pure

postulation, an assertion that a certain created representation

is perfectly adequate, which gave the primary qualities their

privileged status. Compare E. Husserl The Crisis of the European

Sciences and Transcendental Phenomenology (tr. D. Carr; Evanston:

Northwestern University Press, 1970), Part II sect. 9 "Galileo's

mathematization of nature" (espec. pp. 23-41) and Appendix B II

"Idealization and the science of reality -- mathematization of

nature" (espec. pp. 309-310).

(8) Birkhoff, G. and von Neumann, J., "The logic of quantum

mechan ics", Annals of Mathematics 37 (1936), 823-843.

(9) Compare here Sellars and van Brakel on the mani-

fest/scientific image dichotomy: van Brakel does not conflate the

manifest image in the sense of how things seem to us ordinarily

with the postulationally constructed world of the perennial

philosophy, as Sellars does. See especially J. van Brakel,

"Natural kinds and manifest forms of life", Dialectica 46 (1992),

243-263; "Interdiscourse on supervenience relations: the priority

of the manifest image", Synthese, forthcoming; "Empiricicism and

the Manifest Image", ms. 1995.

(10) I cannot except phenomenology from this charge; Husserl

urged us to go back to the things themselves in phenomenological

analysis, but his Platonism was crucial involved in shaping that

analysis.

(11) The sense of "reducible" can in fact not be too strict;

it does not mean that the old excluded descriptions will turn out

to be logically deducible from the new scientific descriptions.

Both Feyerabend and Kuhn's more realistic description of what has

been touted as reduction in the sciences, and leger-de-main with

such ideas as supervenience, functionalism, the intentional

stance, or instrumentalism, give us clues as to 'acceptable'

weakening of the claim.

(12) Nancy Cartwright, "Fundamentalism vs the patchwork of

laws", ms. 1995.

(13) The example, and the point, is not my own: see M.

Heidegger, History of the Concept of Time: Prolegomena (tr. T.

Kisiel, Indianapo lis: University of Indiana Press, 1985), Ch. 2

sect. 5.c.alpha, p. 38.

(14) Compare Feyerabend's distinction between the

characteristic and interpretation of a language in Ch. 2 of his

Realism, Rationalism, and Scientific Method (Philosophical Papers

vol. 1. Cambridge: Cambridge University Press, 1981). It does

not seem to me that we can rest easy with his discussion; there

is not enough there to really speak of a theory, only a sketch

for a theory.

(15) I do take it, contrary to some epistemologists, that the

very point of forming a set of full beliefs (on whatever subject)

is to have a single (therefore consistent, coherent) view (of

that subject). But we do so on specific subjects, confronted as

they come, related to "live" problems-for-us, in ways suited to

exactly those problems.

(16) Again: contrasted with 'local' reconstructions, whether of

large parts of our past or small parts of our present -- such as

logical reconstructions of classical physics or of population

genetics.

================END OF FILE============================