AGAINST TRANSCENDENTAL EMPIRICISM

Bas C. van Fraassen

doc: agains08.93

AGAINST TRANSCENDENTAL EMPIRICISM

Bas C. van Fraassen

pg. sect.

2 1. What empiricism cannot be

3 1.1 What is a philosophical position?

4 1.2 The role of empiricist dogma

7 1.3 First argument against dogma

10 1.4 Second argument against dogma

14 1.5 Prospects for an alternative

16 2. Husserl on the demarcation of science and philosophy

16 2.1 Natural vs. philosophical thinking

21 2.2 Orientations and intentional correlates

22 2.3 Presupposition versus orientation

28 3. Empiricism revisited

29 3.1 Attitude not justified by beliefs

30 3.2 Attitudes as involving belief

32 3.3 The central attitude: science as ideal

36 4.4 Detachment and conditionality in belief

What is empiricism? There can be no authoritative answer to any such question. A historian of philosophy can at best try to call what is common to philosophers who either identified themselves, or have traditionally been identified, as empiricists. But what has set those philosophers apart from others, and especially from those whom they criticized, may not be captured in common views or doctrines. The historian may, in trying to fix the label, rely tacitly on a view of what philosophical positions are and how they are to be identified. Finally, it is typical of philosophers who decide to range themselves under some pre-existing banner ("empiricism", "pragmatism", "phenomenology") to change the very philosophy they take on, as much as did their historical heroes in their day. I will here try to give a sustained argument about what empiricism cannot be, and then enter upon a tentative exploration of what it should be (taken to be).

1. What empiricism cannot be

As I proceed I shall bring in piecemeal some of the characteristics that have characterized those episodes in the history of philosophy which I identify as paradigmatic for empiricism. That is clearly a biased way of proceeding. That I want to be an empiricist is intimately connected with my preference for certain philosophical moves or themes as against some others. When I report characteristics of empiricism, I will include only those I endorse, and neglect what I regard as common errors that past empiricists have fallen into. The goal is to develop a philosophical position which will share what is admirable in past versions of empiricism and which is feasible today.

1.1 What is a philosophical position?

To be an empiricist is to take a certain kind of philosophical position. But what is that? Typical examples that come to mind at once (nominalism vs. realism with respect to universals, dualism vs. monism on the mind-body question) are clearly views of what we and the world are like. They say that certain kinds of beings are real (universals, mind) and how they are related to each other. To have such a view then is to believe something. If we take this as the definitive cue we arrive at the following (meta-)philosophical principle:

Not just any statement X+ will fit here of course. To believe that there are flying saucers does not qualify as taking a philosophical position. But perhaps the belief that there are monads or universals or alternative possible worlds does so qualify.

If Principle ZERO is correct -- and this is a big question, the moment the principle has been formulated -- then it must yield a corollary for empiricism: there must be some statement E+ such that

(NE) To be an empiricist = to believe that E+

I have given this corollary the mnemonic name "(NE)" for "Naive Empiricism." That reveals at once, of course, that I do not regard it as obviously correct (though naiveté does not disqualify one in philosophy). By implication then, I do not advance Principle ZERO as obviously correct. We need to see where all this leads us. the question we face is: what statement could possibly play the role of E+ in (NE)?

1.2 The role of empiricist dogma

If there is such a statement as E+ it is the dogma that sums up empiricism. But what role would it have to play then?

Here I must draw on my view of the history of empiricism. It has often enough been pointed out that in Aristotle's reaction to Plato, we have an empiricist turn, though Aristotle was by no means an empiricist all told. The nominalists at Oxford and Paris in the 14th century, who destroyed medieval Aristotelianism from within, are rightly seen as the beginning of what became 17th and 18th century empiricism in Britain and France. Kant was not an empiricist, but his critique of all past metaphysics was, as I see it, an improvement on the empiricist critique of rationalism. It was an improvement in part because it identified the metaphysics involved in British empiricist, which fell equal prey to that critique.

It would be possible to continue in the same vein: what stands out in these attributions of empiricist aspects of various philosophies is that they are based not on dogma but on a critique. What is similar is the form of critique, which calls us back to experience from the enmeshing webs of theoretical reason. That is not at all the same as the theorizing about experience which typically accompanied it (and on which all these philosophers differed from each other). Therefore, if we were to advance some candidate for the role of E+, it would have to furnish the basis for the critique which the empiricist rebel aimed at their targets in metaphysics.

At this point I can go no further without advancing some opinion about the nature of those targets and the form of critique. (This may be more revealing about my own philosophical intuitions and predictions than about the history of philosophy, for as a philosopher I look at that history with a very selective eye.) The targets I take to be forms of metaphysics which (a) give absolute primacy to demands for explanation, and (b) are satisfied with explanations-by-postulate, i.e. explanations which postulate the reality of certain entities or aspects of the world which were not already evident in experience. The empiricist critiques I see as correspondingly involving (a) a rejection of demands for explanation at certain crucial points, and (b) a strong dissatisfaction with explanations (even if called for) which proceed by postulation.

Suppose that, in a philosophical way, I do not understand ethics or science or religion. It might be one thing to take me by the hand and lead me into relevant experience. That might allow me to acquire a deeper sense of insight into those aspects of human existence. It would be quite another thing -- and to the empiricists of little or no value -- to postulate that there are certain entities or realms of being about which ethics (or science, or religion) tells us a true story. Yet that is what philosophers have often tended to do: to 'explain' ethics by the contention that ethical principles are just the (putative) truths about Values, scientific theories the putative true summary of the Laws of Nature, and religious doctrines the putative true description of a divine, extra-mundane reality. Such philosophical accounts tend to be backed up with the assertion that unless we can think of the relevant text as purporting to be a true story, there is no explaining or understanding the subject at all. The empiricist response is to deny firstly the value of any such 'explanation', and secondly the reasons anyone might have for thinking it to be true, and then furthermore to reject the legitimacy and appropriateness of that demand for explanation itself.

If we now suppose that empiricism consists in belief in a certain empiricist dogma E+, then E+ must be the basis for this response. It appears to follow that E+ must entail either that the postulated entities do not exist or that we cannot have any information about them if they do -- and in addition, I suppose, contradict the claim that the subject is unintelligible if we do not believe in their existence. So we must now inquire whether any statement E+ can both play this role and be itself believable for the empiricist.

1.3 First argument against dogma

The literature does disclose candidates for the role of empiricist dogma, at least by way of slogans that carry the promise of eventual replacement by full-fledged philosophical theories. One such slogan which I have used myself is

Here "information" is intended in its "success" or "endorsing" sense. In a broad sense, soothsayers and the yellow press give us information, as do dreams and indeed any statement which is not a tautology. The narrow sense here intended appears when we say that of course we do not regard what soothsayers, the yellow press, and dreams say as sources of information about the world.

Such a slogan cannot be attacked or defended as it stands; it is at best a promising stand-in for something more precise. I will try to argue that there is, however, no way to turn (*) into a statement which will play the required role as empiricist dogma and simultaneously be acceptable to the empiricist. For my first argument I want to draw on our common knowledge of the logical empiricist writings in the first half of this century to disclose some of empiricism's implicit ideals.

Empirical science has always, for empiricists, been the paradigm of epistemic rationality. This gives them a picture of the ideal rational believer, whose epistemic life resembles that of science itself under ideal circumstances. As is well-known, the required distinction between the scientifically-significant/ cognitively-meaningful and the remainder (to be classed with superstition and metaphysics) proved extremely difficult to draw in general. In practice, however, there is often little doubt about particular cases. Let us introduce a name and short description for people who would instantiate this ideal of rationality:

This is not a complete definition, since its meaning hinges on that of "scientific", "observe", and "test." Like (*), it is more or less a promise or stand-in of something envisaged as more precise. Appealing to our common understanding, it certainly remains vague; but perhaps no more vague than "bald people", "ethnic group", or "socialist."

Being SOBER wouldn't make one an empiricist, or even a philosopher, since this SOBRIETY does not require critical reflection. Conversely, an empiricist need not be SOBER though he must admire this SOBRIETY, or regard it as an ideal. From this it follows that if an empiricist lists (*) -- or some precisation thereof -- as his dogma, then it must be something which can be believed by all SOBER people. Indeed, the belief in (*) must then in some way be the basis for the ideal status of this SOBRIETY.

So belief in (*) would at the very least not disqualify one from being SOBER. In that case, if it is not a tautology then it must belong to he class of scientifically or empirically respectable statements described by (a)-(d). (If it is a tautology then everyone is an empiricist, even the noxious metaphysician. I assume we can discount that possibility.)

But now a new factor comes into play. One of the great virtues empiricists claim for scientific rationality is exactly this: disagreement does not make one unscientific. If any statement is scientifically respectable, and it is not a tautology, then so is its opposite. So if one SOBER person believes (*), and his sister disbelieves (*) then she may be SOBER too.

We conclude therefore that if everything is well with (*), then there can be SOBER people who believe the opposite. This is a picturesque way of explaining that the epistemic rationality admired by the empiricist does not preclude denial of (*).

But then, how could (*) provide the basis for the empiricist critique? Its denial is compatible with what that critique leaves standing even under conditions of ideal rationality. So it counts.

This argument is so general that we should conclude, it seems, that there cannot be an empiricist dogma, in the sense of a statement E+ such that (i) to be an empiricist = to believe that E+; (ii) belief in E+ provides the basis for the empiricist critique of metaphysics; (iii) E+ is not itself a victim of that critique. The basis for my argument were the contentions, common at least to the logical positivist and logical empiricist writings of this century, that there is a distinguishable class of respectable candidates for rational belief, and that direct contraries to those candidates are also respectable candidates for rational belief. Since this sums up a good part of the empiricist admiration for scientific rationality, I think the conclusion is inescapable.

1.4 Second argument against dogma

In the first argument I did rely on an empiricist contention that some empiricists too might challenge. That is the distinction between the respectable candidates for belief (found in science and in the reports of observation we ourselves make or endorse) and the propositions disqualified from such candidacy (as found in noxious metaphysics and engendered by pseudo-problems in philosophy). This distinction has an unfortunate history, since the verification criterion of meaningfulness and other attempts at demarcation of science (or of the cognitively significant) were one and all failures. This history was intertwined with the equally problematic attempt to distinguish between pure observation language and theoretical or theory-infected discourse.

But with all this granted, we still must agree that if (*), or some precisation thereof, qualifies as empiricist dogma, then it must survive challenges of the sort which are typical of empiricist critique. Suppose a soi-disant empiricist asserts

I do not want to ask him how he knows this to be true. He has the right to believe this, and take his stand as a (*)-believer, with the proclamation that he is an empiricist and that (*) fills the role of the empiricist dogma (E+ in formulation (NE) above, which was the corollary to Principle ZERO). But I do certainly want to raise the question: is (*) true? Is it really the case?

On empiricist grounds, how is this question to be answered? To play the role of empiricist dogma, (*) cannot be a tautology -- and the empiricist regards nothing as a priori except what logic can prove. So (*) must be, if true, on a posteriori truth. Even the empiricist may hold that there are facts about this world which are epistemically inaccessible, that is, which it is in principle impossible for science to disclose. But I take it that if (*) belongs to those inaccessible propositions, then the empiricist is automatically precluded from making it the basis for his critique. For then s/he must regard views directly contrary to (*) as rational as well.

In the preceding paragraph I have once more drawn on what I take to be common knowledge about empiricism. As I mentioned before, this amounts not so much to a historical claim (though it could easily be illustrated from historical sources), as a selection of what I endorse in empiricism. If we accept all this as correctly characterizing empiricism, then (*) needs to qualify as a scientific hypothesis, accessible to scientific investigation. The empiricist can then believe it, with the rider that he regards it as an empirical hypothesis, which future empirical inquiry may confirm or disconfirm. (He also believes then that disconfirmation will not happen, or at least would not happen in the long run under ideal conditions of inquiry.)

But I submit that, if we make (*) precise in a way that qualifies it as a topic for scientific investigation, then it will not be able to play the role of empiricist dogma. Specifically, the precise version will either be so restricted in its domain and applicability that it cannot function as the basis for empiricist critiques of metaphysics, or else it will be such that any attempt at an empirical test will be question begging.

What follows here will only be a summary of the argument. Consider firstly how (*) could be disconfirmed. This could happen in two ways. The first is that we find that experience fails to give information. Imagine for example that we put a subject, Peter, in a case with elephants, and when he comes out his opinions about elephants do not fit the facts any better than before. The second is that some information appears to be derived from a source other than experience. For example, we put Peter into a sensory deprivation tank and he emerges with new opinions about elephants which fit the facts much better than did his prior opinion.

In both cases the disconfirmation is illusory, unless we construed (*) in a way the empiricist did not intend. That a certain specific kind of experience (in a cage, say) gives no information about a specific subject (elephants, for example) does not contradict any conceivable empiricist dogma. The finding itself was information derived from experience (being in an elephant cage does not always improve opinion about elephants). Secondly, the idea of a source of information is not reducible to mere change in opinion in some fortuitously fortunate way. When Peter emerges from the sensory deprivation tank with new opinions about elephants, neither he nor his listeners have any new information about elephants by scientific standards. If such experiments, plus auxiliary investigation, reveal a correlation between the sensory deprivation tank dreams or fantasies and the zoological facts then that correlation is the new information found -- but its source is the observations of Peter's posterior state and of elephants, and the collation of those two sets of observations.

Secondly, consider how (*) might be confirmed. Suppose that we find for a large variety of topics X and a large diversity of human subjects Y that, in some carefully demarcated sense, experience of X (by Y and possibly others) is the one and only source of information concerning X for person Y. Can we generalize upon this to support the unrestricted thesis (*)? Here we must think about how this evidence was found. If this was a scientific inquiry, then the findings were based on reports of experience by the investigators. Moreover those investigators were required, by the scientific community, to insert no data not derived from observations in reproducible experimental situations. Therefore the method followed was explicitly designed so as to throw no light on whether (*) holds in the sub-domain formed by this scientific inquiry itself. The inquiry itself may be fine and scientifically respectable, but not as an inquiry into (*) construed with unrestricted domain.

1.5 Prospects for an alternative

Both lines of argument have led to the same dilemma. When we have a candidate, such as (*), for the role of (+) in

(NE) to be an empiricist = to believe that (+)

then we find that it cannot be simultaneously construed as a respectable belief for the empiricist's ideal rational believer, and also as a basis for the empiricist's critiques of metaphysics. Hence nothing can play that role. There is as far as I can see only one way out of this dilemma: we must deny Principle ZERO as a characterization of philosophical positions ueberhaupt, and insist that they can take a different form.

The problem with naive empiricism appears to lie in the very idea of (NE) itself, which entails that to be an empiricist must consist in believing some statement about what the world is like. But is there any other sort of alternative? I think that we have throughout, though implicitly, been dealing with philosophical positions which cannot be captured in dogmas. In characterizing the forms of metaphysics which empiricist attack, I emphasized the demand for explanation and satisfaction with certain kinds of explanation. For empiricists I listed rejection of explanation demands, dissatisfaction and disvaluing of explanation by postulate. Moreover I listed the empiricists' calling us back to experience, their rebellion against theory, their ideals of epistemic rationality, what they regard as having significance, their admiration for science and the virtue they see in an idea of rationality that does not bar disagreement. Notice that not a single one of these factors is a belief. The attitudes that appear in these lists are to some extent epistemic and to some extent evaluative, and perhaps some involve or require belief for their own coherence. But they are not equatable with beliefs. Implicitly, then, we have already been relying on a view of philosophy which belies Principle ZERO.

So here is a radical proposal: a philosophical position can consist in something other than a belief in what the world is like. Taking the empiricist's attitude toward science rather than his or her beliefs about it as the more crucial characteristic, we are then led to the suggestion: the alternative to Principle ZERO is that a philosophical position can consist in a stance (attitude, commitment, approach). Such a stance can of course be expressed, and may involve or presuppose some beliefs as well, but cannot be simply equated with having beliefs.

2. Husserl on the demarcation of science and philosophy

The denial of Principle ZERO engenders the suggestion that philosophical positions can take the alternative form of attitudes or stances taken rather than dogmas believed. This idea is certainly linked -- at least for me -- with Husserl. I refer here to the discussions of what he calls the natural attitude or natural orientation (Einstellung, a term also used for the tuning of a radio or focussing of a lens). That we have been to some extent repeating his statement of the problmatique is certain.

2.1 Natural vs. philosophical thinking

Husserl's outline notes for The Idea of Phenomenology begin:

In the first lecture itself he elaborates this: What is taken for granted in natural thinking is the possibility of cognition. Constantly busy producing results, advancing from discovery to discovery in newer and newer branches of science, natural thinking finds no occasion to raise the question of the possibility of cognition as such. That does not mean that science ignores cognition as a subject of inquiry. Cognition appears as a subject within the scope of science; in fact it does so in two ways:

In our terms: the cognitive sciences (which form part of empirical science) comprise both psychology on the one hand, and formal syntax, semantics, and logic on the other.

So the philosophical questions which Husserl has in mind here are not the questions answered either by psychology or by logic or linguistics or studies in artificial intelligence. They are questions which neither arise nor are addressed in a scientific context. If forcibly introduced there, they change the context of discussion, alter what is taken for granted. They invite the discussants to reflect on exactly what they are doing; and that reflective activity is incompatible with the doing, it would displace the questions at issue there so far.

We are today quite familiar with a different view of philosophy, namely that it is part of science. On that other view we can also see this phenomenon of displacement of scientific questions by philosophical questions. However, the latter are just seen as scientific questions with different topics of concern. Suppose for instance we find a psychologist interested in learning theory, who is running rats in a maze. We invite him to reflect on the methodology followed in his laboratory. Rather than interrupt his own research, he hands the question to a cognitive psychologist, who proceeds to sit in as an observer and studies the rat running research. After a while this cognitive psychologist produces a model of information processing which, she claims, fits that laboratory research. We raise questions about the mathematical structure of that model, and ask about the general family of models which she regarded as appropriate candidates here. Rather than interrupt her own research she gives the question to a mathematical psychologist...; and so forth. Quine's article "Epistemology Naturalized" is in effect the proposal that the fictional we, who kept raising "more fundamental" questions in the above scenario, is engaged in the sole activity legitimate for philosophy. But this we was raising scientific questions. Each question was a factual question, to be answered through further empirical or logico-mathematical research.

Are there other questions at all? Husserl's phrasing is unfortunate here. For he might be read as raising very traditional sceptical questions, when he concludes with:

It is quite clear however that Husserl is not proposing a slide back into hoary disputes over the reality and knowability of the external world.

The results of logical and empirical investigation look like answers to philosophical questions, at first blush, because the linguistic forms are the same:

There is something conceptually skew in this exchange. Realizing that can easily lead one into a philosophical quagmire, like a search for synthetic a priori propositions whose assumptions underlie all empirical inquiry. But it needn't. The realization of how it is conceptually skew can be followed instead by the reflection that the question holds up, for critical reflection, our attitude toward empirical science (which treats experience and only experience as source of information) and its status for us. That attitude, whatever it is, is unquestioned and unscrutinized in the contexts of projects carried out under its aegis.

Questions are sometimes posed in the very form which they are intended to sabotage or subvert, by showing the limits of all questions of that form. "But are worldly possessions really worth anything?" If taken at face value, that can be answered with "Well, my house is worth a quarter of a million, my car about ten thousand,..." But that misses the point.

2.2 Orientations and intentional correlates

It is possible to be stubborn and construe any philosophical question as a factual question within the scope of science. If I ask "What is art?" you may offer me books by an art historian, a sociologist, and a psychologist. Each of these begins by delimiting quite precisely its own topic and scope of inquiry, laying down a demarcation of relevant vs. irrelevant questions. The relevant questions (though different for these three authors) are factual and are appropriate subjects for empirical and logico-mathematical inquiry. But the question may have been intended differently. It may for example have been meant as a request to help me arrive at a settled attitude toward art, informed by an understanding of many of its aspects (which indeed requires a good deal of factual knowledge and personal acquaintance), which will then go some ways toward defining my relationship with the world I live in.

Husserl writes as if there are only two possible orientations, distinguished by the fact that in one, the possibility of cognition is taken for granted and in the other not. In existentialist phenomenology of the early post-war years, more than two different orientations are distinguished by their intentional correlates. The intentional correlate of such an orientation is a world, in the sense in which we speak of the world of (or, described by) physics. However, the connotation of the phrase "the world of physics" is too theoretical. When in that orientation, we are immersed in it: we live in that world, its structure is the structure manifest to us in our experience, we desire and manipulate its objects, it provides the objective context for praxis. "The world of the physicist" might be more accurate. Other orientations correspond to the ways of seeing and living in the world of a clinical psychologist, or artist or aesthete, or perhaps, a religious or someone thoroughly politicized in a certain ideology. Of course these worlds are abstractions, structures to be discerned in experience. Not even the most radical physicalist can pretend that the world depicted by physics is all there is to what he or she is personally experiencing. On the other hand, the use of such terms as "the world of the physicist" or "the world of the aesthete" to denote the intentional correlate of certain orientations can also mislead. It is at best a temporary surrogate for the phenomenological analysis needed to disclose those structures in experience.

2.3 Presupposition versus orientation [stance]

There is an interpretation of what Husserl means that would bring it to naught. He says that in the natural stance, the fact that cognition is possible is taken for granted. Suppose this means:

(a) to be in the natural orientation is to have the

presupposition that cognition is possible

In that case his view implies that the proposition that cognition is possible is, so to say, a hidden axiom of all natural thinking. The difference between being in the philosophical and in the natural orientation is just that of thinking under different assumptions.

Could that be the right construal of Husserl's distinction? Notice how it aligns with the construal of philosophy in Principle ZERO: to take a philosophical position is to adopt a belief. The category of belief is simply widened to the less committal attitude of assumption, with the provision that assumptions can be suspended (and perhaps temporarily replaced by rival assumptions, in the mode of supposition). In each 'overall' orientation (natural, philosophical) we would then see many subspecies formed in exactly the same way, by the adoption or supposition of more specific assumptions.

This construal of Husserl's distinction is indeed an option, but a very disappointing one. It brings us right back to the meta-philosophical views that gave us Principle ZERO and the naive empiricism of (NE). It is, in other words, not an option that will lead us to a viable form of empiricism. If we did take that option -- let us call it the naturalizing option, for that is how I regard it -- we would formulate an "empiricist" view of what science is by adding to the general assumption that cognition is possible a specific one along the lines of (*):

In that case we have come to the disconcerting conclusion that (*) is a presupposition, and hence hidden axiom, of all empirical science. And this would mean that to be in the natural orientation (or indeed, to be a scientist) is to be in the very predicament in which we located naive empiricism. For the whole of empirical science would then rest on a belief whose truth is in principle, and on pain of vicious circularity, beyond the reach of empirical inquiry.

But the reasoning that leads us from Husserl's text to the naturalizing option proceeds by several dubious rhetorical moves. The first is to introduce the notion of presupposition to capture what is meant by the idea that in the natural attitude, the possibility of cognition is taken for granted. In other words, the attitude is identified with a certain epistemic relation (persuasively called "presupposition") to a putative factual proposition about how we are related to the world. The second move is to assimilate that relation of presupposition to assumption and supposition, thus reducing the entire matter to very familiar stuff. The conceptual autonomy of such notions as attitude, orientation, stance is thereby lost or nullified. The better option will be to preserve that autonomy.

After insisting on this logical or conceptual independence of two clusters of terms/concepts, I do want to raise a delicate issue about their connection. I call the issue delicate, because on the one hand it threatens to subvert or sabotage our project, but on the other hand I know very well that I cannot discuss it adequately here. For it is related to other issues in epistemology which I can here only indicate obliquely.

The issue is this. Empirical science does allow as data, ultimately, only those drawn from experience. That is a crucial part of the empirical, scientific method. But you cannot adopt this method, and have confidence in it, without believing that this is just the right thing to do. And that requires that you believe (*). So (it seems) whatever else we are going to do about this, we have to admit that science does involve that belief, whatever subtle distinction we are going to make about presupposition, orientation, etcetera.

What this threatens is an end run around any manoeuvres designed to replace naive empiricism with a more viable, sophisticated variant. For the argument purports to show that science involves or even rests on the belief that the naive empiricist would take as dogma but which -- as we have seen -- empiricism cannot regard as a candidate for rational belief. The objection amounts to: a scientist cannot rationally rely on the scientific method without believing in its adequacy, but any formulation of that belief will be designated as either trivial, or scientifically deniable, or noxiously metaphysical by the empiricist. This problem has really been hanging over us from the beginning. If we can't put it aside, it will be no use trying to make something of Husserl's demarcation without placing metaphysical assertions at the foundations of science.

I want to answer this: to adopt a method (wholeheartedly) requires on pain of incoherence that you have no belief entailing its inadequacy (except for purposes you don't adopt it for!). This limitation on what you cannot believe does have some implications for what your opinion should be -- it is not merely negative. But the opinions that are logically forced on us, in this way, have no content beyond what they are forced by. So the scientist's confidence in his method adds no content to the confidence (concerning specific purely empirical propositions) which he reaches by means of that method. To put it briefly: the objection we face tries to capitalize on a logical point, but no logic can lead from what a scientist believes qua scientist to something that is beyond the reach of science.

It will be objected that even if adoption of or commitment to a method cannot be equated with having the belief that this method is adequate, nevertheless that belief is involved therein. Other attitudes besides belief can involve belief, even inextricably. The problem for empiricism will persist if some belief is involved which empiricism regards as neither a logical tautology nor empirically significant.

In my view, we are not dealing here with belief but a necessary illusion of belief. To point to what I have in mind here (I can do no better now), let me mention the "preface paradox." It may be tempting to think that reason and prudence require the author of a book to say in his preface that some of what follows is false. If he does that he will no longer be asserting the body of the book, but only displaying it and claiming that most of it is true. If he does want to assert it -- and he may be perfectly warranted in doing so, without having apodictic certainty that it is correct -- he is in a quandary. For if he says in his preface, honestly but not tactfully "(I believe that) (unlike in other books) all of this is true" he will sound incredibly arrogant. He should not add this statement, but the logical point remains that he cannot assert anything at odds with this statement while asserting the body of the book. The only confidence this entails on his part, however, is the rightfully earned confidence he has in each statement he wrote in the body of the book. The paradox results from the philosophical tendency to squash pragmatic implications into semantic molds. The idea that adoption of a method must involve a substantial belief (in the method's adequacy) of the very sort that the method was designed to scrutinize critically, derives in my view from a very similar philosophical confusion.

3. Empiricism revisited

It remains now to identify the empiricist stance, to isolate those attitudes that characterize empiricist philosophical positions. Best if I could do it in twenty-five words or less. But that I cannot do. I conceive of this task as follows. Central to empiricism is a certain attitude toward empirical science. But attitudes involve concepts in some way, and are inconceivable without them; they also involve beliefs in some way and are not formulable without them. Hence spelling out this central attitude involves also the articulation of how science is to be conceived.

There must be some leeway: there are undoubtedly varieties of empiricism, predicated on variations in this conception of science. In addition there must be leeway for some variance in the attitude toward science as conceived. Crucial will be the question: just how much of science, which aspects, are valued so highly as to set an ideal for practical and theoretical reason in general? How much of actual science bears out this ideal and what is to be ascribed to historical, cultural, and sociological accident? Most of all: where is even this ideal to be left behind as no longer offering adequate or relevant guidance? Empiricism is not meant to be scientism or idolatry of science; it entails a critique of scientism in the name of science. I will briefly address each of these points.

3.1 Attitudes not justified by beliefs

There is a certain bias toward Reason in most philosophy: attitudes must be justified, their justification must be by something other than attitudes, so attitudes must derive from beliefs -- what else? Isn't the alternative purely subjective preference? How fortunate for us that this question has an exact parallel in sceptical challenges to belief itself! That should undermine its appeal. If a philosopher does accept that attitudes are legitimate or rational only if justified by beliefs, he or she lands immediately in the fact/value problem. For the belief says that things are thus or so, while to justify an attitude it must establish that this attitude is better than or superior to its rivals. The attempt to meet this problem with the assertion that the relevant belief will be one which is (based on, derivative from) genuine insight into value, yields an immediate similar dilemma. Is the content of this insight a factual proposition (about Values perhaps)? Or is having an insight into value the same as valuing, having an evaluative attitude (which is in some sense correct)? Obviously we land in a circle either way. The only way out is to deny the initial bias: rationality does not require our attitudes to be justified by beliefs.

3.2 Attitudes as involving belief

At the same time, attitudes and beliefs are inextricably involved with each other. I cannot admire a person's honesty unless I believe that he is telling the truth, nor can I admire it if I believe that the consequences will come at great cost to others and none to him. It is to be expected therefore that if we isolate those attitudes which characterize empiricism, we will find them inextricably involved with certain beliefs. Since I take it that the attitude toward science is central to empiricism, the centrally involved belief must be a view of what science is.

Since I have addressed this question elsewhere, and since it can only be answered by the very lengthy process of developing an entire empiricistically palatable philosophy of science, I shall not directly address it here. But I want to address two subsidiary questions. One is how we should approach this subject of attitudes involving beliefs. The other is what exactly is believed, in holding a view as to what science is.

One of the worst legacies, to my mind, of eighteenth century British empiricism, is the Humean belief-desire psychology and other quasi-mechanistic attempts to reduce psychological discourse. Even when it is admitted that such discourse is irreducible, and cannot be translated or replaced without loss into e.g. physicalist idiom, the reductionist instinct persists. What of 'internal' reduction? Can all psychological concepts be reduced to logical constructions from just a few -- just belief and desire perhaps? These are legitimate philosophical questions. The damage occurs not when we ask whether such a reduction is possible, but when we let the hope or assumption of its possibility infect our thinking before it has been demonstrated. If, for example, we implicitly require investigations in either ethics or aesthetics to proceed within a certain such simplified psychology, or tacitly impose that requirement as a touchstone of intelligibility, we are in danger of laying waste entire philosophical regions of inquiry. I do not believe that any such policy is an essential ingredient of philosophy in general or of analytic philosophy, much less of empiricism. I must admit, however, that many philosophers have wittingly or unwittingly participated in such conceptual scorched earth tactics.

I remark on this as a plea for investigation of the relations between attitudes and beliefs (as well as goals, intentions, commitments, forms of opinion other than belief) and so forth, unconstrained by premature simplification. My plea extends to the treatment of what science is. This belongs to a specific type of philosophical question that marks large areas of concern: What is science? What is art? What is religion? What is law? Each such question arises also, in similar form but with different intent, within the special sciences: sociology, psychology, anthropology, history. The game metaphor tends to set off the philosophical intent of such questions, though not perfectly: what is the game of science? What is winning and losing, what incurs penalties, what are admissible/good/disastrous strategies? What is the point of the game? What is the aim pursued, what is the criterion of success? These questions are all posed in "folk-psychological" discourse, irreducible to anything devoid of the concept of person, and I think, equally irreducible to simplified questions about individual and communal beliefs and desires alone.

In short, I expect that a very rich psychological idiom will be needed to answer the questions: "how are beliefs involved in attitudes?" and "what is the content of the involved beliefs?" This contrasts sharply with the idea that an empiricist position might be identifiable with a belief to the effect that there is a certain purely semantic relation between scientific theories and facts (or between scientific models and the empirical world).

3.3 The central attitude: science as ideal

For empiricism, science is the paradigm of human rationality; empiricism advocates scientific method as the guide to life. Thus baldly stated, it appears to have two terrible corollaries. The first is that the empiricist stance is just what Husserl calls the natural orientation, and hence the death of philosophy. The second is that what is advocated implies intellectual submission to actual science, which always has been, is now, and undoubtedly always will be very imperfect, limited, Procrustean. So understood, empiricism would be a sacrifice of the intellect and empiricists the running dogs of scientific imperialism.

But that is not so. Being truly scientific necessarily involves a skeptical attitude toward extant science, rejection of "scientism", and refusal to allow any intellectual scheme to keep us pinned and wriggling on the wall. Science has not advanced through submission to its previous generations. The scientific method is the skeptical method, with one difference: it allows (unlike the classical skeptic) that it is also rationally permitted at any point to lay to rest doubts one cannot disprove, and to "bet" on (accept, commit oneself to) a theory which goes infinitely far beyond any evidence we could have. But in this, the intellect is not sacrificed, and rational inquiry is not replaced by blind dogma, provided the step taken is acknowledged as such. That means: to let such a step display itself in the light of day, and not to pretend that it was compelled (as opposed to: permitted) by reason.

This requires the possibility of a certain detachment, of simultaneously having convictions while capable of standing back and assessing them critically. Is that possible at all? Perhaps I have disarmed the second of what I called the two apparent terrible corollaries; but what of the first? In the natural orientation, questions as to the very possibility of cognition, or as to presuppositions of one's current questions (and survey of possible answers to those questions), are absent. The terminology betrays us here, for it harks back to Husserl's dichotomy. There is in fact a large panorama of possible orientations, in each of which some presuppositions are investigable, but every one of which appears blind to certain presuppositions when viewed from within another orientation. Husserl pointed, for simplicity perhaps, to two extreme poles, both impossible: the consciousness so immersed in things that it cannot be troubled by pre-conditions of cognition, and the opposite consciousness so philosophically free that its inquiry is presuppositionless. In actuality, we are never at either extreme. Here as elsewhere, Neurath's image applies of sailors rebuilding their ship at sea.

There is another pressing need at this point. We need to spell out just what an empiricist can believe about science, and I shall address that in the next sub-section. But meanwhile, related to the doubts and misgivings presently at issue is that science triumphant would replace our present belief-structure/world-picture with something truly unpalatable. Though Goethe put this in Mephistopheles' mouth, who of us is not ready with "Grey, dear friend is all theory, but ever green grows life's golden tree"? Or, to shift the words of an English thinker from its original target to one more threatening now:

In the scientific orientation, I approach everything scientifically. If I advocate that approach, do I advocate that when in love, I should consciously keep in mind all of Masters and Johnson's questions while on a walk with my beloved -- and reflect on the gravitational constant as the falling peach blossom petals fill the air with the poignancy of their dying life?

No, that is ludicrous. The insinuation here is that a stance is not limited, that a consciousness could just be a single stance graven in stone. What one learns, apprehends, comes to see and understand during that walk is real, and indeed, it is not among the deliverances of scientifically disciplined experience. The right advocacy of the scientific orientation is to advocate it only against its direct rivals -- the rival approaches to getting the same information about the world by consulting superstition, spirits of the dead, metaphysical argument, innate ideas, or the comforts of explanation and other intellectual indulgences, where these go beyond the test of experience. Love, whether of another or Other, is not such a rival.

3.4 Detachment and conditionality in belief

I have left till last something that I regard as a serious research problem. In describing what empiricism admires in science I have emphasized the ideal of an epistemic attitude which is indeed belief but incompatible with dogmatism. If the empiricist is right then acceptance in science, of theories and hypotheses, does involve belief, but undogmatic belief. The beliefs are held, but with a certain detachment. Is such detachment possible? This question masks a paradox that has, in my view, seriously undermined past attempts to develop an empiricist epistemology. The important and outstanding research problem I see is to fashion a conception of belief which makes this empiricist admiration for science intelligible and tenable.

In my own view of science, acceptance of a scientific theory involves not the belief that this theory is true, but only that it is empirically adequate ("saves the phenomena"). Less than full acceptance involves similarly attenuated opinion as to the theory's empirical adequacy. But these opinions still go very far beyond the evidence we have. They involve many beliefs about observable facts elsewhere and elsewhen. Such beliefs will equally be hostage to the fortunes of future experience. Here is the empiricist's admired attitude: with respect to such beliefs, I must display a certain amor fati. If they become very unlikely in view of new experience then (this must be my resolve) I shall not hold on to them at all costs, but revise my position.

The resolve to allow all one's beliefs to be hostage to the fortunes of future experience, empiricism sees as crucial and central to scientific method. Moreover, empiricism holds this up as an ideal for us with respect to all our beliefs and opinions (and indirectly, through their vulnerability to changes in opinion, also our attitudes generally). But how could this sort of resolve possibly be anything more than empty posturing?

Here 2 is surely a promise (or better, it expresses a resolve or decision) as well as a prediction. But what is the cost of this promise to the speaker? In 1. we see by implication that this speaker fully believes that he will never see a white crow. As a prediction, 2. is vacuous and as a promise it is therefore cost-free, to this speaker. Hence if 2. claims a paramount virtue, it is nothing but empty posturing. It is quite like the promise "If I become a hippopotamus, I will be gentle."

But this indictment rests on a picture of belief and opinion that is too simple-minded. I submit that our straightforward opinions ("I believe that X", "It seems very likely to me that Y") are but he tip of the iceberg of our epistemic position. Nine-tenths of this iceberg consists in what we believe under suppositions, including those contrary to our beliefs tout court. The same conditionality attaches to all our attitudes, resolves, decisions, and intentions in a similar way.

Rather than proceed with this as a general topic in epistemology, I will immediately illustrate this with the beliefs about what science is that are involved in the empiricist attitude toward science. Empiricism certainly includes a belief that the actual cultural phenomenon everyone refers to as science involves centrally theoretical activity whose most important criterion of success is accordance with the data. Suppose now that I, who have both this attitude and that factual belief, am deprived of the latter. This could happen if sociologists of science were to convince me that this criterion is systematically overridden by, and does not override, other criteria of success. I would no longer be able to say that empirical adequacy is the point of the game of science actually played (and so-called) in our culture. One option for me would be to say that today we have no genuine science (and perhaps we never did). Instead, I take it, my resolve (as I see it today) requires that I relinquish my belief about science, and re-evaluate my attitude: perhaps it should still be one of admiration, perhaps not.

Would this mean then that I would give up my philosophical position? In a strict sense, yes, of course. But what this hypothetical scenario draws more attention to is the layered, hierarchical structure of both beliefs and attitudes. I can suppose that some of my beliefs are false, and then consult myself to see what follows for me, and what my attitude would be, under that supposition. To some extent, my present position must be identified with the implicit answers I have to questions raised under suppositions contrary to what I hold now. I have conditional beliefs and other attitudes as well as simple ones. In some scenarios only a very little of my present position is demolished, under others quite a lot.

To continue the above imaginary scenario, consider two options. I may react with: I value science so highly and for so many reasons independent of (my belief in) its search for empirically adequate theories, that I would take whatever the sociologists revealed to me as the true paradigm of rational inquiry. Alternatively I may react with: my paradigm of rational inquiry is science as I presently conceive of it (which subordinates all other criteria to "fit" with respect to the observable phenomena). Therefore I would then cease to say that science embodies rational inquiry (even imperfectly), and maintain that it should not have its common, current role of guiding factual opinion. In each case, I have maintained the more basic part of my position; but my position turned out to be quite different in these two cases. The difference appeared only when I confronted certain hypothetical questions.

In fact, of course, I take it that the second option reveals that the position I had was a form of empiricism. What is maintained here is a certain attitude, expressed by saying that I regard a certain approach to factual questions as paradigmatically rational.

- - - - - - Quantum Mechanics: An Empiricist View. Oxford: Oxford University Press, 1991.