output.txt

1
Theory Change and the Logic of Enquiry:
New Bearings in Philosophy of Science

Anglo-American philosophy of science has tended to define itself squarely against the kinds of (so-called) metaphysical approach that have characterised (so-called) continental philosophy in the line of descent from Husserl. Indeed, Husserl's project of phenomenological enquiry was the target of criticism by Fregeand later by Gilbert Rylewhich pretty much set the agenda for subsequent debate.1 That project seemed to them just a form of illdisguised 'psychologism', one that purported to address issues of truth, validity, rational warrant, and so forth, but which fell far short of the logical rigour attained by thinkers in the other (analytic) tradition. Thus Husserl might claimlike Descartes and Kant before himto be raising questions about the a priori forms of human knowledge and experience, forms that were given (necessarily presupposed) in every possible act of cognition.2 Moreover, he might claim to have advanced beyond Kant in distinguishing more clearly between formal and transcendental logic, or judgements whose necessity followed from the ground rules of this or that logically binding system of thought, and judgements that resulted from a rigorous reflection on the genesis and structure of human understanding in general.3 However, these claims counted for little with Husserl's critics in the other (i.e., post-Fregean) 'analytical' camp. What they chiefly objected to in Husserl's project was the approach via thoughts and ideas 'in the mind' of some perceiving or reasoning subject, even though Husserl was very often at pains to reject any merely empirical (or psychologistic) construal of his claims. To their way of thinking, all this talk about 'transcendental' truth- and validity-conditions was just another variant of the bad old Cartesian-Kantian retreat to consciousness as the last court of appeal in epistemological matters. Only by rejecting that entire line of thoughtthat is to say, by adopting a strictly analytical or logicosemantic approachcould philosophy at last break free of its attachment to naive, subject-centred, or 'metaphysical' notions of meaning and truth.
More recently some analytic philosophers, Michael Dummett among them, have begun to question this doxastic account and to offer a more nuanced appraisal of the differences between Frege and Husserl.4 Nevertheless, there is

still a marked tendency to suppose that, wherever such differences show up, it will always be a matter of Frege's having pointed the best (most rigorous and logically adequate) way forward and of Husserl's having slippeddespite all his strenuous protests to the contraryinto some form of argument that unwittingly reveals his residual subjectivist approach. Now it seems to me that these belated, rather grudging gestures of atonement are partly a result of certain emergent problems within the analytic tradition, in particular those raised by Quine in his famous attack on the two 'last dogmas' or Camap-style logical empiricism.
5 That is to say, they have to do with the impossibilityas Quine sees itof maintaining any version of the logical-empiricist dichotomy between 'matters of fact' and 'truths of reason', or synthetic and analytic orders of statement. This distinction was crucial to the logical-empiricist programme and also, in a somewhat less doctrinaire form, to the whole enterprise of analytical philosophy in the wake of Frege and Russell. Whether Quine actually succeeded in giving it the coup de grâce is still very much an open question, not least (as I shall argue) in view of his failure to offer any convincing alternative approach to issues of truth, meaning, and interpretation.
So there is a case for taking a different view of the grounds for dispute between the two philosophical traditions. This revisionist approach would go much further than Dummett in acknowledging both the importance of Husserl's project and the various problems that analytic philosophy has encountered in its attempt to enforce the Fregean veto on talk of consciousness, intentionality, ideas, eidetic structures, the noetic/noematic distinction, and so forth. In so doing it would raise very different objections to logical empiricism from those put forward by Quine and other thinkers who continue to endorse that veto. After all, Quine is insistent that philosophy of language should adopt a strictly extensionalist semantics which finds no room for such 'opaque' (intensional) items as beliefs or propositional attitudes. Thus, for instance, while it is jointly true that 'W. V. Quine is a Harvard philosopher and author of the essay ''Two Dogmas of Empiricism"', there is no equivalent conjunctive truth-value which attaches to that statement when preceded by the phrase 'Rita believes that. . . .' For she might, as it happens, be in error about various facts, among them his academic calling or his institutional affiliation or his having written that particular essay, or perhaps about the name of whoever it was that answered to one or more of those identifying descriptions. (Maybe she thinks the essay was written by Hilary Putnam or some other well-known Harvard philosopher.) In short, such statements fail to meet the test of substitutivity salva veritate for termssuch as 'Quine' and 'the author of "Two Dogmas of Empiricism"'which pick out an identical referent and are therefore extensionally equivalent. Hence also his argument against the possibility of quantifying into modal contexts, a practise whichon Quine's submissioncreates all manner of insoluble problems, such as

the breakdown of logical equivalence between '9 is necessarily greater than 7' and 'The number of planets is necessarily greater than 7'.
6
Thus Quine has basically the same reason for objecting on the one hand to intensional (belief-related) approaches in philosophical semantics and on the other to modal-logical talk of necessity, possibility, truth across various (more or less epistemically inter-accessible) 'worlds', and so forth. Where they both run into trouble is in failing to secure the functional equivalence of statementsthose containing names or definite descriptionswhose truth-values must be covariant if we are to uphold (as Quine thinks we must) the extensionalist doctrine that terms with the selfsame referent should always be interchangeable salva veritate in every context of usage. Nevertheless, it is hard to see how this argument can possibly work given his relentless attack, in 'Two Dogmas', on the various co-implicated notions (of synonymy, analyticity, truth-by-definition, etc.) which in Quine's view reduce to so many items of otiose circular reasoning. Some would argue on the contrary that such reasoning can have a certain ampliative function, that is, a capacity for drawing out the sense and proper usage of those terms, despite their apparent circularity when treated from a strictly logical point of view.7 Then again, other critics of QuineRuth Barcan Marcus among themhave argued that the problem is not so much with quantifying into modal or intensional contexts but rather with providing an adequate extensionalist theory of meaning and truth. On this view Quine's standard objection (i.e., that modal/intensional approaches are 'referentially opaque') can in fact be turned around to claim that the opacity 'lies with Quine's use of such terms as "identity", "true identity", "equality"'.8 Thus, according to Marcus, logical systems should be treated as 'more or less extensional', the criteria in each specific case depending 'on the kinds of contexts and predicates which are prohibited, and the degree to which the relation of identity is equated to weaker forms of equivalence'.9
However, it is not my purpose here to adjudicate these issues within the broad spectrum of current analytical philosophy of language and logic. Rather, it is to make the more general point: that they all take rise from a certain preemptive narrowing of philosophic sights, namely the beliefone shared by Quine and Marcusthat we shall make progress or achieve more clarity in these matters only by eschewing the kinds of 'subjectivist' approach that are taken to characterise work in the other (i.e., post-Husserlian) continental tradition. In what follows I propose to question this belief and suggest that analytic philosophy has much to gain from a closer acquaintance with developments in continental epistemology and philosophy of science. In particular I shall focus on two such developments, both carried on (albeit from a critical-deconstructive standpoint) in the early writings of Jacques Derrida. These are the projects of Husserlian phenomenology and the distinctive approach to issues of scientific knowledge and theory construction adopted by

Gaston Bachelard. I shall argue that they offer a promising way forward from the various problemsespecially those remarked upon by Quinewhich have dogged the legacy of logical empiricism and its various successor movements. From this point of view the dispute between Quine and Marcus as to whether or not one can quantify into modal or intensional contexts is a regional dispute that has arisen in part as a consequence of their both foreclosing on that wider (intentional or consciousness-related) domain of enquiry.
'It seems to me', Marcus writes, 'that much of the discussion these past few years concerning the apparent breakdown of substitutivity principles in intensional contexts and its presumably devastating results for logic and mathematics are [sic] largely terminological'.
10 And again, more pointedly: 'the apparent difficulties of interpreting such [modal/intensional] systems are not genuine, but analogous to a rejection of non-Euclidean geometry because it allows parallel lines to meet'.11 As it happens, both Husserl and Bachelard are much concerned with the processes of thought by which advances come about in fields such as mathematics, geometry, and physics, advances that can only be achievedso they arguethrough a break with hitherto accepted ideas of self-evident, 'commonsense', or a priori truth. What is at issue hereas Marcus brings out very clearlyis the question how far such processes of thought should figure in a properly logical account of the context of justification. For some analytic philosophers, Quine preeminent among them, there is just no place for intensionalist (let alone intentionalist) modes of talk when trying to make logical sense of such matters. For othersincluding Marcusthere is a firm line to be drawn between intentional approaches (which are strictly indispensable if one wishes to deploy modal, counterfactual, or other such causal-explanatory devices) and intentional approaches which supposedly open the way to all manner of psychologistic vagaries. Thus, on her account, logic can perfectly well accommodate most of the items that Quine objects to ('"knows that", "is aware that", and in particular "is necessary that"') without fear of referential opacity. However, it requires that those qualifying phrases be properly construed, that is to say, in accordance with a modal-logical or strictly intensionalist approach which finds no place for such inscrutable matters as the discovery procedures or the processes of thought by which conjectural beliefs are transformed into adequate knowledge.
It is this dimension of enquiry that was firmly closed off by the 'analytic' turn against Husserlian phenomenology and, indeed, against all those later 'continental' developments perceived as not matching up to the highest, most rigorous standards of logical accountability. My case is that Anglo-American philosophy was thereby deprived of some useful epistemo-critical resources, not least for overcoming the various problems bequeathed by logical empiricism, problems acknowledgedin their different waysby Quine and Marcus. More specifically, the two projects of Husserl and Bachelard can be seen

to address crucial issues that were mostly disregarded by thinkers in the mainstream analytical tradition but which have often surfaced to vex that tradition in the form of such internal disputes as the rift over modal/intensional conceptions of meaning, logic, and truth. Hence the lately emergent strain of 'postanalytic' philosophy, conceivedvery often with reference to Quineas an outcome of just these intractable problems in the doctrine of logical empiricism. I have argued elsewhere that this reactive movement of thought is frequently in danger of swinging right across to the opposite extreme, that is, toward a 'strong' hermeneutical approach (derived from Wittgenstein and Heidegger) or a full-blown Rortian 'textualist' position which finds no room for such old-fashioned notions as reality or truth.
12 Indeed, Quine's argument in 'Two Dogmas' is itself a very striking example of the way that a thoroughly physicalist (or naturalised) epistemology can be made to square with an equally thoroughgoing doctrine of meaning-holism, one that takes the entirety of scientific knowledge or belief at any given term as the 'unit of empirical significance'. And from here it is no large step to those varieties of hermeneutic-textualist doctrine according to which interpretation goes 'all the way down', so that any results turned up (say) through 'direct' observation or empirical research must themselves be treated as interpretive constructs, or as making sense only against some background 'horizon' of communal belief. For if indeed, as Quine writes, 'it is nonsense, and the root of much nonsense, to speak of a linguistic component and a factual component in the truth of any individual statement', then equally it is the source of much confusionor so I shall argueto push this case for meaning-holism to the point where 'truth' becomes entirely a matter of contextual definition.13
Thus for Quine, famously, knowledge is 'a man-made fabric which impinges on experience only along the edges', or again, 'a field of force whose boundary conditions are experience' (TDE, 42). At the centre of the fabric are those so-called logical laws of thought (bivalence, noncontradiction, excluded middle) which are taken to possess a priori validity or to function as the very ground rules of rational debate in the physical sciences and elsewhere. However, Quine cautions, we should not be misled into thinking that these rules are absolutely or intrinsically immune from revision. For this is after all just another doomed attempt to uphold the artificial dichotomywhether Humean, Leibnizian, Kantian, Camapian, or whateverbetween contingent 'matters of fact' on the one hand and necessary 'truths of reason' on the other. 'Taken collectively', Quine writes, 'science has its double dependence upon language and experience; but this duality is not significantly traceable into the statements of science taken one by one' (TDE, 42). And again, pushing the argument through: '[t]hat there is such a distinction to be drawn at all is an unempirical dogma of empiricism, a metaphysical article of faith' (37). Rather, what we feel ourselves bound to accept as logical 'laws of thought' should instead be treated as particularly well-entrenched precepts, principles,

or procedural guidelines which have so far proved to have great utility in working some manageable structure into the otherwise chaotic flux of incoming sensory stimuli. Any conflict at the edges of the fabric (i.e., any 'recalcitrant' experience or item of anomalous observation-data) can always be assuaged by making certain adjustments elsewhere in the system or by redistributing predicates and truth-values so as to restore equilibrium. At times this process may go so farunder pressure of conflicting empirical evidencethat it forces a change at or very near the logical heart of the fabric, that is to say, a revision to the hitherto sacrosanct 'laws of thought'. Thus '[a]ny statement can be held true, come what may, if we make drastic enough adjustments elsewhere in the system' (TDE, 43). And the same applies to those physical objects which we tend to think of as existing 'out there' beyond the boundary of the fabric, but which in factQuine arguesshould rather be viewed as so many 'posits' whose being is dependent on our various theories, ontologies, descriptive languages, conceptual schemes, and so forth.
No doubt, instrumentally speaking, 'the conceptual scheme of science' is a useful tool 'for predicting future experience in the light of past experience'. But this should not deceive us into thinking that the objects posited by sciencefrom quarks and gluons, via electrons, atoms, and molecules, to the whole range of middling and large-scale objectsare more or less 'real' according to some ultimate (non-scheme-dependent) standard of objective truth. For what, after all, could such a standard consist in if not our current best notions of 'reality' derived from a mixture of commonsense wisdom plus other, more specialised (and preferably up-to-date) items of scientific knowledge? In which casegiven the presumed nonfinality or the always revisable status of science as we have itthere is simply no distinguishing between various 'posits' (from centaurs and Homer's gods, via numbers and mathematical sets, to brick houses on Elm Street) in point of ultimate reality or truth. Rather, as Quine puts it,
[p] hysical objects are conceptually imported into the situation as convenient intermediariesnot by definition in terms of experience, but simply as irreducible posits comparable, epistemologically, to the gods of Homer. For my part I do, qua lay physicist, believe in physical objects and not in Homer's gods; and I consider it a scientific error to believe otherwise. But in point of epistemological footing the physical objects differ only in degree and not in kind. Both sorts of entities enter our conception only as cultural posits. (TDE, 44)
The same applies to those relatively abstract entities'classes and classes of classes and so on up'which constitute the object-domain of mathematical reasoning, and whose status has long been a topic of dispute among Platonists, conventionalists, intuitionists, and others. For Quine, such disputes must be seen as altogether pointless, presuming as they do that there exists some answersome ultimate, non-scheme-dependent answerto questions

concerning the 'reality' (or otherwise) of numbers, sets, or classes. 'Epistemologically', he writes, 'these are myths on the same footing as physical objects and gods, neither better nor worse except for differences in the degree to which they expedite our dealings with sense experience' (TDE, 45). In which case we had better just acknowledge that there is nothingat any rate nothing 'in the nature of things', whether physical objects, mathematical sets, centaurs, or Homer's godsthat could possibly justify our ranking those objects on a scale (or on different scales) according to their various well-defined orders of real, ideal, hypothetical, fictive, or presumptively mythic existence. Quite simply, they are all 'cultural posits' which 'enter our conception' just insofar as they allow us to construct some plausible framework for relating 'experience' (or sensory stimuli) to the various other theories, posits, and items of belief that make up our current ontological scheme.
Now the really odd thing about all this is that Quine can be a hard-headed physicalist as concerns those sensory 'stimuli' while nonetheless adopting a wholesale contextualist approach with regard to what counts, on some given interpretive scheme, as a candidate real-world object or entity. Indeed, this might be called the third dogma of empiricism, one that supposedly makes short work of the old analytic/synthetic or scheme/content dualism but which opens up a yet more problematic gulf between this or that item of stimulus-response and the overall 'fabric' or system of beliefs wherein such items have to be somehow accommodated. It is herein the space between causal stimulus and cultural 'posit'that a wide range of current antirealist or cultural-relativist doctrines manage to insert their claim, often in terms that avowedly disown any such characterisation. Thus, for instance, Richard Rorty can declare himself a full-fledged 'realist' in the sense that he accepts (1) the existence of a real-world or mind-independent object domain, (2) the fact that this world impinges directly on our nerve ends or stimulus-response mechanisms, and (3) the necessity of defining 'truth' with reference to just this causal link between sensory stimuli and contents of belief. But he can then go on to claim that items (1) to (3) above have absolutely no bearingor exert no ultimate constraintupon the range of interpretive choices that exist with respect to our various theories, descriptions, or world hypotheses. For on this view it is a species of category-mistake to suppose that whatever transpires in the causal interaction between world and mind is in any way decisive as regards the construal to be placed on those various items of stimulus-input. That is to say, there is no passage from a Quinean physicalist (or naturalised) epistemology to an account of how certain stimuli are linked to certain perceptions, how these latter play a role in certain (more or less reliable) observation-sentences, and so on up to the stage of rational belief or theory formation.
14
Thus, as Rorty sees it, there are two different kinds of 'aboutness' relation involved in the ascription of meanings and beliefs to people who can safely

be assumed to interact with their immediate physical environment in the way that Quine describes. One is the externalist viewpoint adopted by a Quinean 'radical interpreter' who ex hypothesi lacks any inside knowledge of what informants may think or believe and who therefore has to work on the basis of observing just such causal interactions, along with their responses (whether active or verbal) to various incoming stimuli. Then there is the kind of process that we ourselves and (presumably) other people engage in when trying to interpret, describe, explain, or rationally justify items of belief. Thus the first is a basically determinist theory having to do with 'an aboutness relation that ties s [the relevant item] to its objects', or which supposes a direct causal link between real-world occurrent states of affairs and beliefs (or belief-dispositions) on the part of real-world situated agents. The second is a quite different sort of theory, one that is essentially normative in character and has to do with 'the inferential relations between our belief that s and our other beliefs'.
15 Now, according to Rorty, it is simply an errorand the source of much epistemological confusionto conflate or assimilate these two kinds of 'aboumess' relation. For the causal theory is just the sort of thing that applies when we are taking the externalist view of how people acquire or manifest various states of belief. It is therefore wholly devoid of normative implications, issuing as it does from a standpoint of studious noncommitment as regards the likely reasons, evidential criteria, or justificatory grounds that those informants might be expected to produce by way of argumentative support. And conversely, there is there nothing in the normative accountthe 'view from inside'that could claim good warrant through a causal appeal to beliefs reliably attained by exposure to the right kinds of physical stimuli. Such beliefs are entirely unconstrained by the aboutness relation which, in Rorty's phrase, 'ties s to its objects'. Rather, they involve a rational (coherence-maximising) method for ensuring that our various beliefs hang together in a broadly satisfactory way, and also for imputing beliefs to other people on the assumption thatmore likely than notthey will be applying the same principle.
What this amounts to is a version of the double-aspect theory, familiar at least since Kant, which requires a careful separation of realms between issues of causal explanation, on the one hand, and issues of rational justification on the other. More recently the theory has been given a new lease on life by Donald Davidson's idea of 'anomalous monism' as a means of resolving the mind-body problem, or at any rate helping to defuse the quarrel between physicalist and mentalist arguments.16 Such problems arise from a failure to perceive that these two conceptions need not get into conflict just as long as one accepts that they involve quite different kinds of understanding, the one aimed toward a causal-explanatory (perhaps a neurophysiological) account of mental processes and events, and the other concerned to make rational sense of our own and other people's acts, intentions, meanings, and beliefs. So,

from this point of view, there is no difficulty about adopting even a strong (e.g., central-state materialist) position on the issue of mind/brain identity while also holding that the other (intentional or belief-related) mode of talk is perfectly validindeed indispensablewhen applied in the right sorts of context. Hence Davidson's theory of 'anomalous monism': monist insofar as it rejects the idea of any radical dualism between mind and body, but anomalous insofar as it rejects any version of the hard-line reductionist argument according to which mental events are mere epiphenomenal illusions brought about by the working of physical laws whose better understanding would leave no room for such otiose 'folk-psychological' talk.
17 Thus we can all be as realist as we like in the external (causal-explanatory) mode and yet leave room for a strong interpretivist or hermeneutic approach when construing people's utterances, thoughts, and intentions on the basis of a Davidsonian 'principle of charity' which brings them out with as many true or rational beliefs as possible.18
In the same way, Rorty suggests, we can surely go along with the realist's case for the existence of an objective, mind-independent, causally efficacious 'external world' which is not just a constructas the incautious relativist might have itout of our various language-games, conceptual schemes, interpretive conventions, or whatever. However, this gives absolutely no grounds for holding that our own or other people's beliefs concerning that world are in any way constrainedor their truth-value decidedby the causal link that 'ties s to its objects'. Rather, such beliefs are arrived at through an ongoing process of interpretive trade-offs or a Quinean pragmatic 'redistribution' of truth-values and predicates. Thus for Rorty, as likewise for Davidson and Quine, there is no sure routeperhaps no route at allfrom a naturalised (causal) epistemology to a theory of rational belief formation that would take due account of this process and the normative values it brings into play. These latter belong to the 'inside view', that which we occupy in our role as self-conscious, reflective subjects for whom the word 'true' is a 'term of praise' applied to beliefs which optimise our sense of overall purpose and coherence. From the externalist viewpoint, conversely, 'true' is a term which properly applies to just those utterances or items of belief which display the right kind of causal history as tracked by a Quinean 'radical interpreter' with access to the relevant information sources but lacking any kind of privileged epistemic warrant. What is more, this restriction is equally in force when we adopt an externalist perspective on our own processes of knowledge acquisition. For in this case also, as Rorty puts it, we should resist 'the urge to coalesce the justificatory and the causal story', since they involve entirely different (incompatible) orders of truth-claim."19 Epistemologically speaking, 'there is nothing more for us to know about our relation to reality than we already know' on the causal account, that is, as a result of simply observing how we and other people respond to certain stimuli under certain ambient condi-

tions.
20 To jump from this to a normative 'story' about what we ought or ought not to believe as a matter of rational inference is just the kind of error that philosophers typically fall into when they fail to take the pragmatist's simple point about the various senses of 'truth'. Thus (Rorty again): 'the understanding you get of how the word "true" works by contemplating the possibility of a Tarskian truth-theory for your language is utterly irrelevant to the satisfaction you get by saying that you know more truths today than you did yesterday, or that truth is great, and will prevail'.21 Least of all should we suppose that the causal sense of the termapplying as it does to the rockbottom level of unmediated stimulus-responsecould possibly figure in a normative or justificatory account of our various holdings-true.
Now it might well be said that Rorty is untypical in pushing so far with the incompatibilist argument that it severs every link between the two kinds of 'aboutness' relation, that which 'ties s to its objects' in a straightforward causal way and that which concerns 'the inferential relations between our belief that s and our other beliefs'.22 Clearly one motive for adopting this line is Rorty's desire to avoid being characterised as a downright antirealist (or cultural relativist) while still leaving maximum scope for his claim that we can 'redescribe' the world in as many different ways as there exist different language-games, 'final vocabularies', metaphors we can live by, and so forth. Nevertheless, Rorty's route to this wished-for denouement is one that follows a path well beaten by other analytic (or 'postanalytic') thinkers, Quine and Davidson among them. In Quine it takes the form of a naturalised or physicalist epistemology, along with a range of interconnected thesessemantic holism, meaning-variance, ontological relativity, the theory-laden character of observation-statements, the underdetermination of theory by evidenceall of which conduce to a radical split between the baseline (causal) and higher-level (normative or interpretive) stages of belief acquisition.23 In Davidson, likewise, the argument works out as a doctrine of direct realism which allows us 'to reestablish unmediated touch with the familiar objects whose antics make our sentences and opinions true or false', joined to an all-purpose Principle of Charity according to which we have no choice but to interpret other people's sayings and beliefs on the assumption that they must be counted 'right in most matters', and hence sufficiently in agreement with us for those sayings and beliefs to be rationally construable.24 So it is that Rorty can enlist Quine and Davidson in support of his case for combining an outlook of sturdy commonsense realism vis-à-vis beliefs and objects-of-belief with the idea of their always being open to creative 'redescription' in line with our current best interests or most favoured cultural values.
From here it is a short step to those varieties of jointly 'postanalytic' and depth-hermeneutical thinking which look to Heidegger for a way beyond the apparent dead endor the seeming trivialityof truth-definitions of the type proposed by thinkers in the mainstream analytical tradition.25 For these

latter reduce either to a Tarskian notion of truth as a purely definitional construct ('"snow is white" is a true statement if and only if snow is white') or else to a crudely empiricist conception according to which the truth-content of sentences, statements, theories, and so on, should best be construed as a matter of 'surface irritations'sensory stimuli of different kindsplus whatever is needed in the way of Quinean across-the-board pragmatic adjustment. So it is not hard to see why some recent commentators have thought this a somewhat impoverished account of what goes on when we actually engage in the process of trying to figure out meanings and beliefs. One response is to urge that this whole way of thinking is hopelessly in thrall to an outworn epistemological paradigm which conceives 'truth' in terms of correspondence (of accuracy, factual content, propositional form, or whatever), and which thus remains closed to that other, more authentic or primordial dimension where truth is a matter of unconcealment, of the bringing-to-light of a wisdom concealed by the accretions of so-called Western metaphysics.
26 On this view philosophy has much to gain by the turn toward Heidegger and depth-hermeneutics, even if temperedas in Rorty's caseby a measure of pragmatist scepticism as regards Heidegger's more portentous or echt-ontological pronouncements. To another way of thinking, no such drastic remedies are called for, but rather a careful reengagement with issues of truth, meaning, and interpretation that have so far not been adequately treated by philosophers in the broadly analytic camp. On this view, therefore, we had much better stick with the resources of conceptual analysis and critique, whatever the problems hitherto encountered in applying those resources to particular questions of epistemology and interpretation theory.
Thus, for instance, Simon Evnine remarks of Davidson that he seems to have 'two quite different projects . . . [o]n the one hand a causal, explanatory project . . . [o]n the other an interpretative, hermeneutic project', and that he fails to bring them together in any convincing or coherent way.27 This problem itself has two aspects, the one concerning Davidson's idea of 'anomalous monism' (which in effect leaves it entirely unexplained just how we are to conceive the mind-body issue or the question of intentional versus physicalist accounts of human action), the other concerning the Davidsonian Principle of Charity and just how it relates to his otherwise thoroughly causal or determinist theory of belief acquisition. For clearly there is a sense in which people (ourselves and others) can indeed be wrong about a whole range of thingswhether motives for action, grounds for scientific belief, items of everyday commonsense 'knowledge', interpretive ascriptions of meaning or belief, sensory perceptions, inferential reasoning on this or that item of evidencewhere the Principle of Charity would bring them out right merely as a matter of a priori warrant for the claim to understand them on the basis of a shared rational or truth-optimising strategy. In such cases what is needed, rather, is a jointly causal and normative epistemology which explains (1) how such

errors come about, (2) why we are justified in viewing them as errors, and (3) what should count as an adequate reason for proposing some alternativebetter informed or scientifically more adequateapproach to the issue.
28 Otherwise Charity will tend to work out as a kind of all-purpose hermeneutic pretext for counting people right (or rationally justified) no matter what they believe just in order to ensure that we don't lose touch and find ourselves forced into notions of radical paradigm incommensurability or meaning-variance across different languages, cultures, 'conceptual schemes', and so forth.
This is of course where Davidson takes issue with Quine, Kuhn, Feyerabend, Whorf, and other proponents of a relativist doctrine with just that unfortunate consequence.29 But it is not at all clear that Davidsonian Charity is much better placed in this respect. For, if consistently applied, then it will always come out either by counting other people 'right' on their own terms as a matter of general (truth-optimising) principle or by simply reconstruing their utterances on the 'charitable' basis that they must agree with us in most of what they say and believe, and hence that we can safely (justifiably) impute meanings and beliefs in large-scale accordance with our own ideas of rational, truth-preserving warrant. In which case, again, the way is wide open to a strong-redescriptivist (Rorty-style) argument that interpretation in some sense goes 'all the way down', since there is nothing in the bottom-line causal theory of how beliefs hook up with objects-of-belief that could possibly serve to fix or individuate higher-level items of belief-content, propositional attitude, ontological commitment, and so on. Thus the hermeneutic 'turn' in some recent readings of Davidson is one that takes off from precisely what Evnine remarks (in the passage cited above) as the split between his causal-explanatory project and a charity-based theory of belief attribution whose normative or justificatory criteria are completely uncoupled from any such appeal to our causal transactions with the world. One very welcome consequenceso Rorty arguesis to cut down the role of epistemology (or theory of knowledge) to a point where it is finally squeezed out between the claims of a naturalised causal etiology of beliefs and a hermeneutic theory premised on the notions of meaning-holism, Davidsonian Charity, and the pragmatist idea of 'truth' as whatever merits that name by our present cultural-interpretive lights.30 Of course Rorty has no problems in reconciling this with Tarski's proposallikewise interpreted via Davidsonthat a theory of truth for any given language should amount to just a formal recursive device which can be seen as simply cancelling through for all practical purposes, and which therefore (arguably) lacks any kind of substantive ontological or epistemological content. So, on this account, the main drift of recent postanalytic philosophy has been to drive home the hermeneutic message: that epistemology is a hopelessly failed endeavour, since 'knowledge' and 'truth' amount to no more than our currently favoured ways of interpreting whatever is given at the basic

level of unmediated (causally determined) stimulus-response. In which case clearly the sole remaining task for philosophy of mind and knowledge is to follow Rorty's lead and devote its main efforts to deconstructing all those grand theories erected by philosophersfrom Descartes and Kant to the logical empiricistswho have laboured under the delusion that epistemology had other, more important or constructive work to do.
31
As I have said, this message has not gone unchallenged among those who take a more sanguine view of the epistemological enterprise but who reject the idea of a strict separation between externalist (causal-explanatory) theories, on the one hand, and normative (reason-based or justificatory) arguments on the other. Thus, according to William Child, 'the idea that a normative, interpretative story is itself a causal story is essential for understanding the form of reason explanation'. Moreover, 'it is also essential for understanding the simple realist thought, that it is possible for us to have thoughts about, and knowledge of, ''an objective public world which is not of our own making"'.32 This last phrase is quoted from Davidson and may be taken to mark the latter's dissent from any version of the strong-descriptivist (Rorty-style) argument that would find no room for an 'objective public world' except as a kind of noumenal posit, a reality that exists (by very definition) beyond whatever we can know or believe concerning it, and which otherwise enters our conceptions merely as a useful shared point of reference in the languagegame of commonsense realism. All the same, it is far from clear that Davidsonian 'objectivity' amounts to anything more robust or secure than Rorty's pragmatist recasting of 'realism' as whatever is (currently and contingently) 'good in the way of belief'. For the effect of decoupling the causal and the normative conceptions of truth is to render the one entirely devoid of justificatory principles or grounds, while the latter floats free in a generalised willingness to interpret beliefs on a wholesale principle of charity which makes no distinction between those arrived at by reliable (knowledge-conducive) means and those adopted in line with existing assumptions or habits of thought. Thus perhaps it is the case, as Davidson writes, that a propositional (or belief-based) psychology 'cannot be, or be incorporated in, a closed science' of the type aspired to by physical theory and those branches of cognitive psychology that take their model from the physicalist paradigm of law-governed causal explanation.33 But the problems are just as great if one swings right across to the opposite (hermeneutic or strong-descriptivist) extreme, or indeed if one adopts a positionlike Davidson's 'anomalous monism'which on principle refuses to perceive any possible conflict between beliefs brought about by causal interaction with the world and beliefs entertained as a matter of maximal coherence with our own or other people's convictions, world-views, or truth-conditional hypotheses. For this gives Rorty just the argument he needs for declaring a complete severance between 'truth = what-

ever shows up in our unmediated causal interactions with the world' and 'truth = whatever we believe as a matter of cultural preference plus least resistance to incoming stimuli under this or that favoured (belief-preserving) interpretation'.
My point is that Davidson's doctrine of anomalous monism in effect gives us the worst of both worlds: a reductively physicalist account of causal explanation and a theory of interpretive 'charity' which equates truth (or rationality) with a precept of across-the-board optimal adjustment between various ontological posits and items of belief. Thus, in Quine's words, '[e]ach man is given a scientific heritage plus a continuing barrage of sensory stimulation, and the considerations which guide him in warping his scientific heritage to fit his continuing sensory promptings are, where rational, pragmatic'.
34 What drops completely out of sight on this account is the process of 'rectification and critique' that plays such a crucial role in Bachelard's epistemology, and which involves a very different understanding of the relation between sensory 'promptings', on the one hand, and the 'heritage' of received scientific thought on the other.35 For Bachelard, this process cannot be reduced (as Quine would have it) to a matter of merely pragmatic adjustment, one that redistributes predicates and truth-values across the entire 'fabric' of existing beliefs so as to preserve maximum coherence with least disruption to cherished items 'Conservatism figures in such choices', Quine remarks, 'as does the quest for simplicity'.36 But in that case it becomes hard to explain how science could ever have confronted a challenge to the orthodox wisdom of its time, or why 'normal' science (in Kuhn's terminology) should ever have found itself thrust into a phase of 'revolutionary' turmoil.37 For there would always exist a whole range of alternative possibilities for reconstruing any item of 'recalcitrant' evidence so that it agreed with the body of accepted scientific lore. Or again, one could always tweak a few threads here and there in the overall fabric so as to accommodate the rogue item while avoiding any challenge to deeply entrenched beliefs.
Such would be the 'rational' line of least resistance as Quine describes it in 'Two Dogmas of Empiricism', that is, the most pragmatically acceptable way of coping with anomalous data. However, this argument fails to account for those decisive 'epistemological breaks' which, according to Bachelard, mark the transition from metaphor to concept, or from the stage of prescientific image and 'reverie' to the stage of adequately theorised scientific knowledge. Above all, it evades the central issue of just why scientists should at times have broken so far with consensual norms as to postulate theoriessuch as Copernican astronomy or quantum mechanicswhich at first seemed completely at odds with all standards of rational scientific warrant. Now one might think that Quine is well placed to explain such episodes since it follows from his twin theses of ontological relativity and radical meaning-variance that everything is in principle always open to revision, from the construal placed

upon incoming sensory stimuli to those putative logical 'laws of thought' that occupy the centre of the fabric. Indeed, quantum mechanics is Quine's chief example of the way that new developments in the physical sciences (e.g., the wave/particle dualism) may force changes in the ground rules themselves, as for instance by requiring a suspension of excluded middle or bivalent truthvalues. Elsewhere, oddly enough, Quine retracts this claim by arguing that it could never be the best (most rational) option to relinquish those ground rules in response to some set of anomalous or conflicting data.
38 For thenas Aristotle pointed outwe should find ourselves in the unfortunate predicament of endorsing both of two contradictory statements, and thus (by implication) asserting every statement that could ever be made on any conceivable topic.
No doubt there are deep problems with the interpretation of quantum mechanics, problems that have led some physicists, philosophers, and formal logicians to propose an alternative (i.e., nonbivalent or many-valued) logic in order to save scientific appearances.39 However, there is also reason to think that this is a somewhat desperate stopgap remedy adopted for want of any better understanding of events in the quantum domain. My point, more generally, is that none of these issues could ever have arisenfrom the break with 'classical' (pre-1900) ideas about the subatomic structure of matter to the further reaches of current debateif those concerned had simply followed Quine's advice and finessed any conflicts by revising the import of observation-statements or redistributing truth-values wherever required. In that case normal science would continue pretty much undisturbed by the occasional (short-term) problem of adjustment between incoming sensory 'stimuli' and the background 'heritage' of scientific lore. What cannot be explained on Quine's account is the emergence of a theorysuch as quantum mechanicswhich not only transformed the fundamental concepts of physical science but created equally fundamental problems regarding its own interpretation, problems that are still (some hundred years on) just as far from being resolved.40 It is therefore something of an irony that a good many currently à la mode doctrines in philosophy of scienceontological relativity, radical meaning-change, paradigm incommensurability, antirealism in its various formsare assumed to derive support from quantum mechanics, or at least to reflect the generalised uncertainty attaching to quantum phenomena. This is probably one reason for the widespread acceptance of Kuhn's ideas about 'revolutionary' science, not to mention the curious reversal of priorities by which some commentators (Rorty among them) have treated it as just the sort of thing that should always go on if we don't want to be stuck with boring old 'normal' science.41
However, there are several things wrong with the equation 'quantum mechanics = undecidability = truth as relative to whatever framework (system of measurement, paradigm, conceptual scheme, language-game, discourse, etc.) one happens to adopt'. First, quantum mechanics assigns a precise

valuethat given by Heisenberg's constantto the margin of uncertainty obtaining for any given measurement in any given case. Second, it is an error (and one with absurd consequences, as witness the strictly unthinkable predicament of Schrödinger's cat) to extrapolate directly from these paradoxes in the quantum domain to objects and events at the nonquantum (i.e., macrophysical) level.
42 Third, whatever the problems encountered in making sense of such phenomena, they are problems that arise precisely from the effort to stretch our minds around a range of well-attested experimental findingssuch as those of wave/particle dualism or nonlocal interactionwhich are not just artefacts of some chosen paradigm, observation-language, theoretical framework, or whatever. And this despite the fact (also well attested) that the so-called collapse of the wave-packet into one or another state is in some sense produced, or its outcome decided, by the act of observation/measurement.43 For there is no good reasonantirealist prejudice asideto derive the conclusion that quantum 'reality' is whatever we make of it according to this or that interpretive scheme. Rather, it is a matter of allowing for effectslike that of observation on the system observedwhich would themselves have a place in any complete (so far unobtainable) description of the various physical factors involved. In short, these arepace Bohr and his disciplesproblems of a basically epistemological rather than an ontological nature.44 That is to say, it is at least more rational to suppose that they reflect our current limited knowledge of events at the quantum level, and not some ultimate mystery at the heart of things that places such events forever and intrinsically beyond reach of scientific understanding. Thus what decides the truth of any given hypothesisas between, say, the standard Copenhagen view and Bohm-type 'hidden-variables' theoriesis whether or not it provides an accurate description of just those physical events, irrespective of any beliefs that we or others may happen to hold concerning them.45
For the antirealist, conversely, it cannot make sense to posit the existence of 'verification-transcendent' truths, items of fact (of rational warrant or logical necessity) that would render our beliefs determinately true or false even in cases where we didn't know enough to ascertain the correct answer, or where currently available proof-procedures (e.g., in mathematics) were inadequate to the task in hand, or where the evidence gave no means of settling the issue one way or another.46 Thus for some analytic philosophersnotably Michael Dummett'warranted assertability', not truth, is the most that we can justifiably hope for in assessing the content of various statements in mathematics, science, history, and other disciplines of thought.47 Such statements can only be evaluated in terms of our current best methods for bearing them out, whether through processes of mathematico-logical deduction, empirical evidence testing in the physical sciences, or reference to reliable source texts and archives in the case of historical research. What cannot be sustainedso this argument holdsis the notion of truth-values that possess an objective, real-

world, evidence-transcendent, or non-belief-relative status. Still less can the realist have grounds for asserting that we might just be wrong about a whole range of itemsfrom mathematical theories to scientific or historical truthclaimsthat come out true according to our best procedures of verification but which are nonetheless false when measured against the way things stand 'in reality'. For this involves us in the strictly nonsensical claim (at least as the antirealist sees it) that we might have grounds for both asserting some putative truth and at the same time assertingwith reference to some higher, more objective tribunalthat our assertion could always turn out false when subjected to more rigorous examination or when all the evidence was in. Quite simply, there is nothing that can warrant such truth-claims in the first place except their conforming to acceptable standards of logical procedure, disciplined enquiry, observational warrant, investigative scholarship, and so forth. Any appeal beyond those standardssuch as the realist's typical appeal to verification-transcendent truthsis one that by very definition could never be made good within the limits of knowledge or existing proof-procedures at any given time. In which case, the antirealist argues, it is a mere dogma of naive 'metaphysical' realism that we can well do without since in giving it up we avoid all the sceptical problems that result from making this otiose distinction between 'truth' insofar as we can possibly know it, and 'truth' as determined by objective standards quite aside from our present best knowledge. For it is just because realists insist on that distinction that they set themselves impossibly high standards for the verification of belief, and are hencevery oftendriven into postures of reactive sceptical doubt.
48
In Dummett's case the route to this conclusion started out from intuitionism in philosophy of mathematicsthe case that truth-values apply only to theorems for which we possess some adequate (decidable) proof-procedureand then led on to kindred claims with respect to science, history, and other fields of enquiry.49 However, it is not ham to see why quantum mechanicsat least on the standard Copenhagen interpretationhas often been invoked in support of similar antirealist positions. Thus it is argued (1) that quantum phenomena (such as wave/particle dualism or the collapse of the wave-packet) cannot be assigned a determinate value apart from our acts of observation/measurement; (2) that this entails the impossibility of positing a 'real-world' quantum domain beyond such effects of observer interference or outcomes contingent on the kind of experiment one chooses to conduct; and moreover (3), that it may also mean suspending certain axioms of classical logic (such as bivalence or excluded middle, if not perhaps noncontradiction) since these cannot apply to 'undecidable' instances like those mentioned in item (1), above. However, as I have said in connection with Quine, this whole line of argument is oddly out of touch with the actual development of quantum mechanics and the kinds of conceptual challenge it has posed during the past near-century of intensive discussion and research. For the quantum theory

could never have emerged and continued to raise such acute interpretive problems if scientists had actually worked on the premise that nothing was in principle immune from revision, whether our conception of a real-world (mind-independent) physical domain or the most elementary ground rules of logical thought. Had they taken this viewequating what is 'rational' with whatever is pragmatically best (or least vexatious) in the way of beliefthen they would always have reinterpreted any recalcitrant data so as to minimise the extent of their conflict with existing scientific theories. Dummett has his own differences with Quine on these and related issues.
50 But if one puts together his antirealist position and his case for the revisability of logic in response to certain (e.g., mathematical or quantum-theoretical) dilemmas, then it is no great distance to Quine's ideas of ontological relativity and radical meaning-variance. And from here the way is open to yet more extreme Rorty-type arguments for the 'reality'-constitutive powers of language, or Kuhnian/strong-sociological claims for the paradigm-dependent or culture-specific character of scientific theories.
I have argued that one alternative to this whole way of thinking is the approach to epistemology and philosophy of science developed by thinkers in the 'other' (continental) tradition such as Husserl and Bachelard. What this approach has to offer, more specifically, is a way of treating these issues that combines analytical rigour and precision with a phenomenological account of the various emergent conflicts and resistances'obstacles' to thought, as Bachelard describes themwhich have marked every major stage of advance in the modern physical sciences.51 But in so doing it avoids the retreat to any kind of naive psychologism or any version of the 'subject-centred' appeal to Cartesian ideas of truth vouchsafed through privileged epistemic access. Rather, it involves a critique of precisely such ideas since they 'tend toward fulfilment in the realm of the image' and hence block the way toward more adequate modes of scientific conceptualisation.52 This applies especially to those branches of modern sciencesuch as quantum physicswhose conditions of emergence included the earlier break with intuitive or a priori notions of truth in fields like mathematics and non-Euclidean geometry. However, this process can never be treated as a simple fait accompli, a once-for-all passage beyond the naive idea that phenomenal intuitions (in Kant's famous phrase) must be 'brought under' adequate concepts. On the one hand, it involves a constant and repeated questioning of truths that appear self-evident not only from the standpoint of commonsense (nonscientific) belief but also in keeping with received ideas of scientific theory and method. On the other, it cannot be conceived as somehow transpiring in a realm of absolute ideal objectivity cut off from the historically emergent processes of thought that led up to some crucial paradigm shift or (in Bachelard's terms) some decisive 'epistemological break' with previous scientific beliefs. It is herein a region

largely unexplored by rational reconstructions in the analytic stylethat philosophy of science can seek to comprehend the dynamics of scientific theory change and the various motivating pressures and conflicts that brought such changes about.
53
Imre Lakatos made a related point (albeit from a different methodological perspective) when he modified Kant's dictum to read: 'philosophy of science without history of science is empty', while 'history of science without philosophy of science is blind'.54 To which it might be addedwith reference to Husserl and Bachelardthat these two endeavours must at some point be linked by a common concern with the genesis and structure of scientific theories, that is to say, their emergence through a critical-reflective process of thought which always strives toward 'absolute ideal objectivity' yet can never be entirely detached from its origins in certain historically situated acts of consciousness.55 In conjoining the terms 'genesis and structure' I am alluding to Derrida's early essay on Husserl, along with other texts (such as his near book-length introduction to Husserl's Origin of Geometry) which are likewise centrally concerned with issues in epistemology and philosophy of science.56 For those texts stand outcontrary to received 'analytic' opinionas among the most rigorous and searching accounts of the difficulties faced by any attempt, like Husserl's, to reconcile these two ('genetic' and 'structural') approaches while avoiding the twin temptations of naive 'psychologism' on the one hand and empty formalism on the other. When read alongside Derrida's commentary on Bachelard in his essay 'White Mythology' they offer what amounts to a sustained engagement with just those problems that have preoccupied mainstream analytic philosophy of science since the high days of logical empiricism.57
I can best make the point by citing a passage from 'Genesis and Structure' which captures the gist of Derrida's argument concerning the strictly undecidable order of priority between these terms. It seems to me that the passage will bear close scrutiny not only by those who equate deconstruction with a Rorty-style 'textualist' approach but also by those in the analytic camp who consider Husserl to have opened the way to a form of psychologism thinly disguised by his talk of ideality, consciousness-in-general, transcendental (as opposed to empirical) psychology, and the like.
If Husserl gives up the psychological route when confronted by all the difficulties of accounting for a structure of ideal meaning on the basis of a factual genesis, he no less rejects the logicizing conclusion with which his critics wished to corner him. Whether in the then current Platonic or Kantian style, this logicism was preoccupied above all with the autonomy of logical ideality as concerns all consciousness in general, or all concrete and non-formal consciousness. Husserl, for his part, seeks to maintain simultaneously the normative autonomy of logical or mathematical ideality as concerns all factual consciousness, and its original dependence in relation to a subjectivity in general; in general, but concretely. Thus he

had to navigate between the Scylla and Charybdis of logicizing structuralism and psychologistic geneticism (even in the subtle and pernicious form of the 'transcendental psychologism' attributed to Kant). He had to open up a new direction of philosophical attention and permit the discovery of a concrete, but nonempirical, intentionaliry, a 'transcendental experience' which would be 'constitutive', that is, like all intentionality, simultaneously productive and revelatory, active and passive. . . . Husserl will attempt to prepare an access to this common radicality through the diverse 'reductions' which are presented initially as neutralizations of psychological genesis and even of every factual genesis in general. The first phase of phenomenology, in its style and its objects, is structuralist, because first and foremost it seeks to stay clear of psychologism and historicism. But it is not genetic description in general which is disqualified, but only the genetic description which borrows its schemas from naturalism and causalism, and depends upon a science of 'facts' and therefore on an empiricism; and therefore, concludes Husserl, depends upon a relativism incapable of insuring its own truth; therefore, on a scepticism. The transition to the phenomenological attitude is made necessary, thus, by the impotence or philosophical fragility of geneticism when the latter, by means of a positivism which does not understand itself, believes itself capable of enclosure by a 'science-of-facts', whether this be a natural science or a science of the mind. The expression 'worldly genesis' covers the domain of these sciences.
58
I have quoted Derrida's text at considerable length since it brings together all the main lines of argument developed in the course of this chapter. In brief, they have to do with (1) the relation between deconstruction and phenomenology, (2) that between approaches to philosophy of science in the two ('analytic' and 'continental') traditions, and (3) the problems that result when analytic philosophers attempt to exclude all reference to the activity of thoughtthe conscious, reflective, or critical activityconcomitant with advances in scientific knowledge or episodes of scientific theory change. This is what Derrida calls (following Husserl) a 'positivism which does not understand itself', one that 'borrows its schemas from naturalism and causalism', thereby becoming a 'science-of-facts' devoid of epistemological content or warrant. So it is that empiricism opens the way to those varieties of sceptical-relativist doctrine that typically arise when philosophy encounters the absolute limit (the in-built resistance to thought) of naive phenomenalist or sense-data theories. After all, as Derrida remarks elsewhere, 'empiricism has always been determined by philosophy, from Plato to Husserl, as nonphilosophy, as the philosophical pretention to nonphilosophy, the inability to justify oneself'.59
It is not hard to see how this remark might apply to epistemology and philosophy of science in the wake of logical positivism. For when that programme collapsed under the pressure of various sceptical critiques as well as through its own internal contradictionsand when Quine followed up with his famous attack on the two 'last dogmas' of logical empiricismthen the way appeared open to a range of relativist arguments which exploited the sheer inability of any such doctrine to explain or to justify our knowledge

of the growth of scientific knowledge. Some philosophers (Feyerabend chief among them) saw this as a happy deliverance, a liberating break with all those old oppressive notions of scientific truth, method, rational procedure, inference to the best explanation, and so forth.
60 Others againsuch as Quine, Kuhn, Davidson, Rorty (at times), and the later Putnamwent various ways around in attempting to formulate an approach that would hold the line against out-and-out relativism while retaining a commitment to what Derrida calls the 'schemas of naturalism and causalism'.61 That is, they all took it that the best (indeed the only) way forward for epistemology and philosophy of science was one that treated knowledge as a direct product of our causal interaction with the world, with those objects and events whose impingement on our nerve ends and the rest of our sensory-cognitive apparatus was enough to keep us reliably informed.
Hence Davidson's well-known advice on how to throw out the third (i.e., the residual Quinean) dogma of empiricism and thus block the way to those sceptical worries that resulted from misplaced talk of 'paradigms', 'frameworks', 'conceptual schemes', and the like. 'In giving up the dualism of scheme and world', he writes, 'we do not give up the world, but re-establish unmediated touch with the familiar objects whose antics make our sentences and opinions true or false'.62 One can find many passages to similar effect in Putnam, Rorty, and indeed (ironically) in Quine himself, all recommending the adoption of a thoroughly naturalised epistemology which is meant to put us directly in touch with reality by cutting out the needless epistemological detour through concepts, categories, schemas, a priori intuitions, forms of representation, and suchlike. But this whole line of argument is fatally weakened by its reliance on a crudely empiricist idea of sense-data (or phenomenal experience) as somehow immediately given to the mind through a process of passive registration where nothing is required save a direct response to incoming sensory stimuli. Thus one need not be a card-carrying Hegelianor even a disciple of Husserlto note how implausible is Davidson's claim that the truth or falsehood of our various sentences (along with their role in our belief systems, commonsense ontologies, scientific theories, etc.) can be just a matter of 'unmediated' contact between our nerve ends and whatever impinges on them from one moment to the next.
With Quine this leads straight on to the doctrine of ontological relativity since there is nothing in the nature of those bare sensory 'stimuli' that could possibly impose any limit on the range of stimulus-compatible schemes, frameworks, or theories. Logical considerations might narrow it down to some extent; for example, by discounting conflictual hypotheses or theories entailing some clear violation of bivalence or excluded middle. Yet of course Quine also makes the case, in 'Two Dogmas of Empiricism', that these putative ground rules of logic or 'truths of reason' are themselves always open to

pragmatic adjustment under pressure from conflicting evidence, thus once again opening the way to full-scale ontological relativity. In Rorty, as we have seen, the causal theory of belief acquisition goes along with a yet more extreme variety of cultural-relativist thinking, that is, the notion that sensory 'data' do nothing whatsoever to fix or to guide the various interpretations that may be placed on them according to this or that theory, language-game, or currently acceptable (scientific or other) communal form of life. And in late Putnam the idea of 'internal realism'internal to this or that frame of referenceseems scarcely more promising, a Kantianism minus the transcendental subject (but still giving rise to all the problems of scheme/content dualism pointed out by Davidson), and a relativism that dare not speak its name but which lies open to the standard counterarguments rehearsed by philosophers from Plato down.
63
My point about these various dilemmas in the wake of logical empiricism is that they all take rise from what Derrida discernsin the long passage on Husserl cited aboveas the inevitable upshot of a positivist 'science of facts' that rejects any form of phenomenological or epistemo-critical enquiry. When he remarks that this approach 'borrows its schemas from naturalism and causalism', the comment applies with particular force to attemptssuch as those of Quine, Davidson, Putnam, and Rortyto overcome the problems of logical empiricism by simply cutting out all that otiose talk of 'truth', 'correspondence', conceptual adequacy, and so on, and adopting a straightforward physicalist theory of 'unmediated' stimulus-response. However, this solves none of the problems, as can be seen most clearly from their various periodic changes of mind (or, in Rorty's case, his nimble balancing act) concerning the realist/antirealist debate and the associated issue of framework-relativism.64 What leaves them in a quandary on both countsand thus gives Rorty his tactical room for manoeuvreis the narrowing of epistemological sights to a point where only sense-data (or 'stimuli') possess direct veridical warrant, and where everything else (from observation-statements all the way up to high-level theories) must therefore be regarded as framework-relative products of interpretation.
This upshot is manifest in Quine's 'Two Dogmas' and alsowhatever his claim to have transcended such old 'metaphysical' habits of thoughtin Putnam's pragmatist recasting of Kantian themes under the name of 'internal realism'.65 It emerges most poignantly is in certain passages of Kuhn's The Structure of Scientific Revolutions, most of them occurring in the 1969 secondedition postscript but some of them to be found in the original (1961) text. What they reveal is Kuhn's distinct unease with the Quinean (physicalist) theory of belief acquisition and also, closely related to that, his sense that there might be something quite wrong with his (i.e., Kuhn's) line of argument concerning such topics as paradigm change, radical meaning-variance, the

theory-laden character of observation-statements, the underdetermination of theories by evidence, and the idea that scientists somehow quite literally 'live in different worlds' from one paradigm to another. Thus, for instance:
Is sensory experience fixed and neutral? Are theories simply man-made interpretations of given data? The epistemological viewpoint that has most often guided Western philosophy for three centuries dictates an immediate and unequivocal, Yes! In the absence of a developed alternative, I find it impossible to relinquish entirely that viewpoint. Yet it no longer functions effectively, and the attempts to make it do so through the introduction of a neutral language of observation now seem to me hopeless.
66
Compare this with the passage from Derrida on Husserl, cited above, and you will gain some sense of the extraordinary narrowness of historical and intellectual vision that led Kuhn to his doleful conclusion vis-à-vis the prospects for philosophy of science at the time. That vision is confined to just one major line of descent (broadly speaking, from Hume to his modern 'analytic' progeny) and, moreover, to just the kinds of problem thrown up by movements in the wake of logical positivism. Thus Kuhn's way of stating the issueas a matter of imposing 'man-made' theories or interpretations on 'fixed and neutral' sense-datawould appear remarkably crude to anyone possessing a knowledge of developments in the other main tradition of philosophic thought running from Kant to Husserl. He or she would also most likely want to challenge Kuhn's claim that this has been the 'epistemological viewpoint' that has guided Western thought on such topics for the past three centuries, one which leaves no room for any 'developed alternative' despite the hopelessness (as Kuhn sees it) of all attempts to establish a neutral observation-language. For there do exist other, highly developed alternativessome of which I have discussed in the course of this chapterthat avoid the predicament so plaintively evoked in the above passage from Kuhn. That is, they follow Husserl in rejecting the positivist idea of knowledge as a product of raw 'sense-data' which are then somehow linked up with concepts (hypotheses, theories, covering-law statements) through a process of so-called rational reconstruction that bears no relation to the processes of thought by which scientific discoveries actually come about.
This approach was carried on by the logical empiricists who introduced a whole range of further distinctionsfor example, that between first-order (material) and higher-level formalised languages, enshrined in the Carnapian doctrine of semantic ascentwith a view to expounding the structure and logic of scientific theories quite apart from the everyday details of scientific practice.67 But when that doctrine collapsed under sceptical pressure (especially from Quine's 'Two Dogmas') the way appeared open, via meaningholism, to various proposalsKuhn's among themfor relativizing truth to this or that framework of in-place consensus beliefs. Hence his two rhetorical

questions ('Is sensory experience fixed and neutral? Are theories simply manmade interpretations of given data?'), questions that Kuhn clearly regards as inviting a negative response but which he still feels driven to ask for want of any viable alternative approach. However, they will seem less pressingor at any rate not so far beyond hope of an answerif one looks outside the restrictive purview of debates within the analytic tradition, at least as standardly defined. For it will then become apparent that there exists a large body of work, some of it inspired by Husserl or Bachelard, that has engaged these issues at a high level of conceptual rigour and analytical precision while avoiding the various drastic dichotomies bequeathed by logical empiricism. Chief among these is the dichotomy imposed by its account of what constitutes a basic observation-sentence (grounded in empirical 'sense-data') and a properly formulated scientific theory (i.e., one constructed out of just such data through deductively valid or inductively warranted procedures of reasoning on the evidence). It is just this sharply disjunctive view of the relation between evidence and theoryalong with the consequent reactive slide into various forms of positivism, naturalism, scepticism, and relativismthat Husserl diagnosed as a 'crisis' besetting both the physical sciences and the very project of enlightened critical thought.
68
Thus the great task for Husserl, in Derrida's words, was 'to open up a new direction . . . [to] permit the discovery of a concrete, but nonempirical, intentionality, a ''transcendental experience" which would be "constitutive", that is, like all intentionality, simultaneously productive and revelatory, active and passive'.69 By so doing it should yet be possible to reverse the tide of rampant unreflective positivism in the natural sciences and also the backwash that, according to Husserl, had produced the contemporary flight to a range of post-Nietzschean irrationalist philosophies. This task was by its very nature interminable, Derrida thinks, since it involved certain deep-laid antinomies (structure/genesis, concept/intuition, 'active' and 'productive' versus 'passive' or 'revelatory' modes of understanding) which would always at the limit defeat any attempt to reconcile their various conflicting claims. Hence also the vital distinction (as Husserl described it) between exactitude and rigour as aims characteristic of different sciences at different stages or depths of phenomenological enquiry. Thus:
an eidetic descriptive science, such as phenomenology, may be rigorous, but it is necessarily inexactI would rather say 'anexact'duc to no failure on its part. Exactitude is always a product derived from an operation of 'idealization' and 'transition to the limit' which can only concern an abstract moment, an abstract eidetic element (spatiality, for example) of a thing materially determined as an objective body, setting aside, precisely, the other eidetic elements of a body in general. This is why geometry is a 'material' and 'abstract' science. It follows that a 'geometry of experience', a 'mathematics of phenomena' is impossible: this is an 'attempt doomed to miscarry'. This means in particular, for what concerns us here,

that the essences of consciousness, and therefore the essences of 'phenomena' in general, cannot belong to a structure or 'multiplicity' of the mathematical type. Now what is it that characterizes such a multiplicity for Husserl, and at this time? In a word, the possibility of closure . . . What Husserl seeks to underline by means of this comparison between an exact and a morphological science, and what we must retain here, is the principled, essential, and structural impossibility of closing a structural phenomenology.
70
So there is no question of finally resolving these issues in the philosophy of mathematics, geometry, and the physical sciences, a point that Derrida himself underlines by reference to Gödel's Incompleteness Theorem and other related developments concerning the limits of computable or axiomatic proof-procedures. But there is also no question of relapsing into the kind of naive positivist attitude that would treat such issues either as belonging to a realm of abstract mathematical concern beyond its own empirical remit orinsofar as they involve talk of Husserlian 'essences', structures of consciousness, or 'eidetic descriptions'as merely a species of misguided psychologism. For this doctrine goes so far in its rejection of 'metaphysical' talk that it effectively debars any genuine engagement with the processes of thought that accomplished those breaks with natural (commonsense-intuitive) understanding or received scientific ideas.
Even when it first appeared during the 1920s logical positivism was philosophically regressive and oddly out of touch with recent scientific developments. These included the emergence of quantum mechanics with its sizable challengeon whatever construalto the positivist conception of knowledge as acquired through a passive registration of sensory data that could then be cast in propositional form and brought under some suitable set of generalised covering-law statements.71 Small wonder that Kuhn was at a loss as to how this theoryin its later, logical-empiricist guisecould possibly provide an adequate basis for explaining the structure of scientific revolutions. Indeed, his whole approach to that topic is perhaps best viewed as the result of Kuhn's disillusionment with the programme of logical empiricism, along with his failure to conceive or envisage any viable alternative account. Thus the 'structure' that Kuhn rather vaguely refers to is one that seems to exist in an abstract realm quite apart from the particular featuresthe tensions, resistances, conflicts of allegiance, struggles to reconcile theory and observationwhich have marked periods of looming transition or prerevolutionary ferment. It is not so much that Kuhn ignores these factors (in a sense, his book is about little else), but rather that he treats them from a viewpoint that excludes all concern with the processes of thoughtat whatever conscious, unconcious, or preconscious levelthrough which they emerged as an obstacle or challenge to existing scientific beliefs.
In this respect, ironically enough, Kuhn's approach gives rise to something

very like the old logical-empiricist split between history and philosophy of science, or what actually went on in the 'context of discovery' (including the false starts, failed experiments, anomalous data, ambiguous observations, dubious conjectures, and other such messy details) and what should go on in the 'context of justification' when philosophers seek to rationally reconstruct the logic of scientific enquiry.
72 Of course Kuhn rejects this distinction in principle since he sees it as upholding the notion of science as a steady progress from stage to stage in the advancement of scientific knowledge. That notion has to go, he thinks, when we acknowledge the incommensurability of paradigms and the amount of semantic gerrymandering that has to occur before we can presume to rank different theories on a common evaluative scale, or claim to know more (compared with our scientific forebears) about historically shifting concepts such as 'mass', 'gravity', 'motion', 'space', 'time', and so forth. All the same, he still adopts an externalist approach that treats these issues, so to speak, from outside and above. That is, he assumes that they can make sense only from the standpoint of a present-day narrative account which relates various episodes in the history of science without judging their truth-content and without the kind of detailed enquiring-back into their epistemological conditions of emergence that would offer grounds for contrasting them in point of descriptive accuracy, explanatory power, or theoretical refinement. In other words, Kuhn rejects one kind of metahistorical perspectivethat which treats science as an onward and upward march toward standards of truth prescriptively defined by its own evaluative criteriaonly to espouse another such perspective according to which past and present theories must always be treated as relative to (or constructed by) this or that range of paradigm-specific beliefs. This approach may claim the liberal virtueas with the strong sociologistsof not rewriting the entire history of science to date in accordance with our own present notions of truth, rationality, and method.73 But in fact it operates on a kind of reverse-discrimination principle, discounting a whole range of past scientific achievements (and the often strenuous activity of thought which led up to them) as possessing no distinctive claim to validity or truth.
This results from Kuhn's highly generalised idea of what constitutes the 'structure' of scientific revolutions, that is to say, his habit of abstracting away from the particular intellectual contexts in which such revolutions occurred and the particular kinds of conceptual, theoretical, or epistemo-critical advance which brought them about. With the phrase 'epistemo-critical' I am once again referring to Bachelard's project of rationalisme appliqué, an outline of which I have sketched already, and which offers, I would argue, a promising way forward from the dilemmas of both logical empiricism and Kuhnian paradigm-relativism. More specifically, it shows how the process of scientific theory change can be treated from a standpoint that takes due account of the thinking that produced such changes, but which nonetheless avoids any naive

(philosophically and scientifically retrograde) appeal to a priori concepts, primordial intuitions, privileged epistemic access, or whatever. It is just this possibility that Derrida holds out in his reading of Husserl as constantly engaged in a strenuous attempt to navigate between the twin polesthe 'Scylla and Charybdis'of 'logicizing structuralism and psychologistic geneticism'.
74 It is also what Derrida finds most revealing in Husserl's text on 'The Origin of Geometry' where he raises the question: how is it possible that an axiomatic-deductive science, one supposedly grounded in truths self-evident to reason, should yet have developed through various stages of historically datable discovery, from Euclid down to the emergence of alternative (nonEuclidean) geometries?75 In the end Husserl cannot answer this question, caught up as he ison Derrida's accountin a constant shuttling exchange of priorities between the rival claims of 'structure' and 'genesis'. On the one hand, 'it is always something like an opening which will frustrate the structuralist project. What I can never understand, in a structure, is that by means of which it is not closed'.76 On the other hand, any too pronounced swing toward the opposite ('geneticist') pole will always run the risk of falling into a naive psychologism which fails to explain how sciences such as geometry, arithmetic, or physics could ever have accomplished a decisive break with intuition-based or a priori conceptions of space, time, and number.
In Husserl this aporia takes the form of an alternating rhythm, a constant switching of priorities, such that these two projects of thought (roughly speaking, 'structuralism' and 'genetic phenomenology') coexist in a kind of tense but revealing mutual interrogative exchange. In Derrida, it is taken up and developed to a point where those tensions are yet more visible but where they are seen as a strictly inevitable outcome of critical enquiry into the structure and genesis of scientific truth-claims. Here again there is a pointed, by no means merely offhand or casual relevance when Derrida refers to the 'difficulties' posed for Husserl's philosophy by 'certain later developments of axiomatics and by Gödel's discoveries'.77 What is in question is not some vaguely generalised notion of 'undecidability' that would take Gödel's theorem as a pretext for suspending all notions of objective truth, all criteria of right reason or standards of valid argument. Rather, it concerns the surprising fact that every formal systemfor example, that of arithmeticcontains certain strictly undecidable propositions, axioms whose truth is a matter of stipulative warrant but which cannot be proven within the system. As Roger Penrose states it: 'no formal system of sound mathematical rules of proof can ever suffice, even in principle, to establish all the true propositions of ordinary arithmetic'.78
In this case it might be supposed that there exist any number of different arithmetics, each of them equally valid by its own formal criteria, since none of them can be both consistent and complete. However, this is to mistake the import of Gödel's Theorem, according to which mathematical

truththe truth of that theorem includedis 'something absolute, there to be discovered, rather than invented'.
79 In other words, the very fact that mathematical reasoning can establish the incompleteness of its own formal proof-procedures is itself evidence that we possess other sources of knowledge, sources that exceed any possible power of computation or demonstrative proof. What the theorem tells us, Penrose suggests,
can be viewed in a much more positive light, namely that the insights that are available to human mathematiciansindeed, to anyone who can think logically with understanding and imaginationlie beyond anything that can be formalized as a set of rules. Rules can sometimes be a partial substitute for understanding, but they can never replace it entirely.80
In making this case Penrose has his eye on the wider implications of Gödel's theorem, especially as applied to issues in the field of artificial intelligence, where it suggestsin line with the central thesis of his bookthat no computer could ever be programmed to achieve such creative-intuitive insights. However, the main point here is that he shows very plainly why the theorem cannot be construed as somehow 'proving' that arithmetic is an arbitrary construct, that mathematical truths are purely conventional, that there exist any number of possible ways of performing some given calculation, continuing some sequence of natural numbers, or whatever. On the contrary, according to Penrose, 'it is within mathematics that we find the clearest evidence that there must actually be something in our conscious thought processes that eludes computation'.81 For otherwise there could be no explaining how Gödel arrived at his (on the face of it) highly counterintuitive result.
What the theorem thus demonstrates is the existence of truthsits own includedwhich are there to be discovered (not invented) but which lie beyond the reach of any rule-governed, algorithmic, or computable process of thought. It is worth remarking, though Penrose doesn't make the point, that this issue about 'following a rule' (what counts as a rule? which of all the various conceivable rules that might apply in this or that case? how to place a limit on the regress from rules to rules-for-the-application-of-those-rules? etc.) is one that has been raised and most often pushed in a sceptical direction by various philosophers after Wittgenstein.82 Indeed, it is one source of that drift toward forms of cultural-relativist thinking which has been such a prominent feature of recent 'continental' and 'postanalytic' philosophy alike.83 So there is a further significance to Penrose's argument concerning the correct interpretation of Gödel's Incompleteness Theorem. What that argument shows, in brief, is that problems with the notion of 'following a rule'whether genuine or spurious problemsare in any case strictly beside the point when it comes to instances of reasoning (such as Gödel's proof) which pass beyond the limits of computable or rule-governed calculative thought. Thus 'no such system of rules can ever be sufficient to prove even those propo-

sitions of arithmetic whose truth is accessible, in principle, to human intuition and insightwhence human intuition and insight cannot be reduced to any set of rules'.
84
On the standard construal of this problem from Wittgenstein the only resort is to a 'sceptical solution' (Kripke's phrase) which restates the problem in a slightly sharpened form and then proposes that we set it aside by appealing to shared practises or communal procedures within which those rules just are the way we do things as a matter of received (e.g., mathematical) convention. But of course this 'solution' solves precisely nothing if one asks (like Wittgenstein) what ultimate justification there could be for claiming that someone who said '2 + 2 = 5', or who continued the number series '2, 4, 6' with the sequence '11, 16, 29', was in some fairly obvious way failing to grasp the arithmetical point. It seems to me that this is just a pseudoproblem produced by the standard Wittgensteinian move of rejecting all truth- or validity-claims except those sanctioned by the fact of their playing a role in some given 'language-game' or cultural 'form of life'. Nevertheless, there is further support to be had from Gödel's theorem as Pen rose interprets it, namely, his argument that reasoning cannot be entirely rule-governed since we have access to a wide range of truths (in mathematics and elsewhere) which exceed the bounds of computability.
However, the chief point I wish to make is that Derrida's usage of 'undecidability'in his reading of Husserl and elsewhereis concordant with this understanding of the term and not with the kind of loose deployment to be found in some adepts of literary deconstruction, and likewise in the notion of 'postmodern' science entertained by thinkers such as Jean-François Lyotard.85 On this interpretation it amounts to no more than a catch-all substitute for other ideas'uncertainty', 'chaos', 'paralogism', quantum nonlocality, the wave/particle dualism, or the limits of precise measurement, to take just a few from Lyotardall of which supposedly signal the end of science as a rational, truth-seeking enterprise. For Derrida, conversely, the reference to Gödel occurs more than once at a specific juncture in his argument where the undecidability is that which results from a conflict between two strictly indispensable but strictly incompatible logics of enquiry.86 In Husserl's case this conflict has to do with the issue of priority between genesis and structure, an issue that Derrida is unable to resolve (any more than could Husserl) but which he presses to the point where it reveals just why there should exist this deep-laid tension in the project of transcendental phenomenology. Moreover, it is a tension that will always emergeso Derrida argueswhen philosophy reflects on the conditions of possibility for scientific knowledge in general and on the various stages of historical development that have marked its progress to date. For the logical empiricists these were to be treated as two quite separate disciplines, the latter concerned with history of science and the so-called context of discovery, while the former pursued the more important task of

reconstructing and testing scientific theories with benefit of rational hindsight, or with respect to their success or failure as measured by standards properly pertaining to the 'context of justification'. As I have said, this dichotomy persists in the work of thinkers like Kuhn, those who reject the logical-empiricist programme but who still deploy the term 'structure' in a generalised sense that scarcely makes contact with the detailed processes of paradigm change or the particular conflicts of theory and evidence that bring such changes about. The result is on the one hand to empty that term of all genuine descriptive-explanatory content and on the other to leave an opening for just the kind of wholesale relativist approach (exemplified by the strong sociology of knowledge) that equates 'truth' with whatever holds good for some de facto, empirically determined community of belief.
Derrida's main concern in his work on Husserl is with the tension between structuralism and phenomenology, each being construed (as I have said) in a sense that goes beyond their normal application to recent European movements of thought. If 'a certain structuralism has always been philosophy's most spontaneous gesture',
87 then equally it is the case that a certain phenomenology has always asserted its claim whenever thinking has turned toward questions regarding the history and genesis of the various sciences. In the context of Derrida's early (pre-1970) work this issue was chiefly engaged by French phenomenology in the wake of Husserl and the emergent structuralist 'sciences of man' which defined themselves squarely against that previous, subject-centred or 'humanist' approach to questions of truth, meaning, and interpretation. These latter of course took their bearings from Saussurean structural linguistics and proposed to extend that pilot science to a whole range of other disciplines, among them anthropology (Lévi-Strauss), narratology (Greimas, Todorov, early Barthes), poetics (Jakobson), psychoanalysis (Lacan), history of ideas (Foucault), and political theory (Althusser).88 What they all had in common was a fierce repudiation of the earlier phenomenological paradigm which accorded priority to the human subject as a locus of thoughts, meanings, expressive intentions, or acts of will that by their very nature exceeded the grasp of any purely structural account. For some thinkers, mainly in the structuralist camp, this conflict of interpretations was such as to allow no possible compromise or via media between the two approaches. For others (Ricoeur and Merleau-Ponty among them) it required a patient and meticulous thinking-through of the various philosophical problems that arose in attempting to reconcile their different claims.89 Where Derrida pursues a distinctive path is in showing that this tension between 'genesis' and 'structure' is one that in principle can never be resolved but which nonetheless opens onto questions of the profoundest import for epistemology, philosophy of science, and the humanistic disciplines. That is to say, those questions will simply not emerge if one adopts either a purebred phenomenological ap-

proach which denies the constitutive role of pregiven 'structures' or concepts, or (conversely) a thoroughgoing structuralist approach that ignores the extent to which thought and language may exceed and transform those preexisting structures or concepts.
Thus it is important to recognisedespite and against one common misreading of his workthat when Derrida 'deconstructs' the discourse of Husserlian phenomenology he does so not in order to discredit its claims but in order to demonstrate the 'absolute and principled necessity' of carrying that project forward in accordance with its own self-critical precepts. The following passage makes this case with particular clarity and force. 'All these [i.e., Husserl's] formulations have been possible', Derrida writes,
thanks to the initial distinction between different and irreducible types of genesis and structure: worldly genesis and transcendental genesis, empirical structure, eidetic structure, and transcendental structure. To ask oneself the following historico-semantic question: 'What does the notion of genesis in general, on whose basis the Husserlian diffraction could come forth and be understood, mean, and what has it always meant? What does the notion of structure in general, on whose basis Husserl operates and operates distinctions between empirical, eidetic, and transcendental dimensions mean, and what has it always meant throughout its displacements? And what is the historico-semantic relationship between genesis and structure in general?' is not only to ask a prior linguistic question. It is to ask the question about the unity of the historical ground on whose basis a transcendental reduction is possible and is motivated by itself. It is to ask the question about the unity of the world from which transcendental freedom releases itself, in order to make the origin of this unity appear.
90
There is much that could be said by way of unpacking this densely argued passage, but I must here content myself with a brief commentary. First, it brings out the sheer distance that separates Derrida's approach to these issues from the 'structuralist' approach manifested in works such as Foucault's Les mots et les choses.91 I choose this particular text for comparisonand with a backward glance to what I said about Kuhn several paragraphs abovebecause it shows very clearly how 'a certain structuralism' affects the very project of historical enquiry into the various forms, stages, and developments of scientific knowledge to date. That Foucault rejected the 'structuralist' tag as applied to his own work is I think more a case of strategic track-covering than a genuine or deep-laid difference of approach. For there are two chief features of Les mots et les choses (and other works of that period) which lend themselves to precisely that description. These are (1) his resolute assault on all notions of the subject, whether the Cartesian 'subject-presumed-to-know' or again, more specifically, the transcendental subject as theorised by philosophers from Kant to Husserl; and (2) his idea that all claims to knowledge or truth must be seen as products of a certain 'structural a priori', one that has nothing whatever to do with any (pseudo)transcendental mode of argument

but which conforms to the dominant order of discourse at any given time and hence undergoes a more or less drastic mutation with every change in that order.
92 No doubt Foucault pushes both arguments much further in a sceptical-relativist direction than would be acceptable to Kuhn or like-minded thinkers in the Anglo-American camp. What they do have in common is a notion of 'structure' that is so devoid of epistemological contentso far detached from the kinds of thought-process involved in particular instances of scientific theory changethat there seems no choice but to give up seeking to explain such episodes and to treat them rather as periodic shifts in the prevailing paradigm or the dominant discursive 'order of things'. In which case there quickly ensue all the well-known problems of meaningvariance and paradigm incommensurability, problems that Kuhn (to his credit) perceived more clearly than Foucault but to which he could provideas I have argued elsewhereno more convincing an answer?93
It is in response to questions such as these that Derrida insists on the continuing necessity of thinking through the genesis/structure relationship as raised to a high point of critical consciousness in Husserl's writings. The various distinctions that he draws from Husserl in the above-cited passage ('worldly genesis' and 'transcendental genesis', 'empirical structure', 'eidetic structure', 'transcendental structure', etc.) are unlikely to carry much weight with philosophers in the Anglo-American tradition. However, I would suggest, it is chiefly for want of such epistemo-critical resources that this tradition gave rise to a doctrinelogical empiricismwhose perceived shortcomings have opened the way to Quinean, Kuhnian, and other varieties of scepticalrelativist thinking. That is to say, it engendered a series of well-known problems that were strictly insoluble on its own terms, among them problems concerning the relation between theory and evidence, the status of empirical or observation data, the validity of generalised (covering-law) statements, and the question as to what should count as scientific progress given the lack of any sure criterion for evaluating theories in point of observational accuracy or conceptual-explanatory power. So it was that Quine could propound his sceptical thesesontological relativity, the underdetermination of theory by evidence, the theory-laden character of observation-statementsas a consequence of just these perceived failings in the logical-empiricist programme. From here it was but a short step to those varieties of full-fledged relativist doctrine (such as the 'strong programme' in sociology of knowledge) that likewise trade on the radical divorce between logicist conceptions of scientific method devoted exclusively to the context of justification, and empirical enquiries into the history of science that expressly reject all normative criteria of truth, reason, and progress. For the latter can perhaps best be seen as a reactive movement of thought, one that embraces the authority of the social (rather than the physical) sciences, and which does so precisely in a gesture of revolt against the narrownessthe lack of any genuine historical

dimensionthat typifies the dominant analytic approach in its rational-reconstructive mode.
94 However, the strong sociologists, like Kuhn before them, can be seen as reverting to just the kind of uncritical positivist outlook that comes of relying on the sheer self-evidence of sociohistorical or factual-documentary data. In short, their approach is still dictated by a version of the logical-empiricist dichotomy, one that rejects the logicist horn of this particular dilemma while scarcely questioning the methods applied and the results turned up by empirical research in the social sciences.
I should not wish to claim that these problems receive a definitive answer in the broadly 'continental' (post-Kantian) tradition of thought that is represented on the one hand by Husserlian phenomenology and on the other by that form of epistemo-critical rationalisme appliqué developed by thinkers such as Bachelard. Indeed, it is a chief point of Derrida's argument that these issues (e.g., that of 'genesis' and 'structure' or the relation between metaphor and concept in the discourse of science and philosophy) cannot be resolved but must always be addressed over again with each new stage in the history of thought. However, this is not to embrace some kind of irrationalist doctrine, any more than Penrose adopts such a view when he takes it (on the evidence of Gödel's Incompleteness Theorem) that there exist mathematical truths and powers of human insight 'beyond formal argument and beyond computable procedures'.95 Derrida is coming at the same issuealbeit from a different philosophical anglewhen he remarks that 'it is always something like an opening that will frustrate the structuralist project'; moreover, that 'what I can never understand, in a structure, is that by means of which it is not closed'. And again, citing some passages from Husserl's Cartesian Meditations:
[t]he results of phenomenology are 'metaphysical, if it be true that ultimate cognitions of being should be called metaphysical. On the other hand, what we have here is anything but metaphysics, in the customary sense with which metaphysics, as "first philosophy", was instituted originally'. 'Phenomenology indeed excludes every naive metaphysics . . . but does not exclude metaphysics as such'. For within the most universal eidos of mental historicity, the conversion of philosophy into phenomenology would be the final degree of differentiation (stage, that is Stufe, structural level or genetic stage).96
Clearly this passage takes aim against the use of 'metaphysics' (or 'metaphysical') as a routine term of abuse, one that has often been applied by analytic philosophers to signify the kind of meaningless talk that they think typical of work in the 'other' (post-Kantian continental) tradition.97 For it is precisely that undifferentiating usagethe failure to distinguish 'naive' metaphysics from 'metaphysics as such'that characterised the discourse of logical positivism and also (in many cases) the thinking of its logical-empiricist heirs and successors. As a result, Anglo-American philosophy entered upon a period of

self-imposed isolation from developments that might have pointed a way beyond some of its more intractable dilemmas.
Now at last there are signsfor instance, in the recent work of Michael Dummettthat analytic philosophers are coming to question this received view, especially as concerns the orthodox charge (handed down by Frege and Ryle) that Husserl's project was really just a form of naive psychologism decked out in spurious 'transcendental' colours.
98 Also, there is a growing recognition that philosophy of science in the broadly analytic mode may yet have something of importance to learn from the critical-rationalist approach of thinkers such as Bachelard and his student Georges Canguilhem.99 Perhaps the next stage in this process will be a reading of Derrida's early texts that acknowledges both their analytic power and their bearing on issues, such as those discussed here, that have been either ignored or inadequately treated by philosophers in the mainstream Anglo-American line of descent. In chapters 3, 4, and 5 I shall offer some further suggestions as to what such a reading might entail. Meanwhilein chapter 2I examine a number of recent analytic (or 'post-analytic') developments that have prompted this awareness of the need to engage with alternative modes of thought.