Datasets:

kothasuhas
/

philosophy-textbooks-16

Dataset card Viewer Files Files and versions Community

Split (1)

train · 12.5k rows

text stringlengths 277 5.29k	textbook_name stringclasses 16 values
on by Hume (1965) and Kant (1987). The ﬁrst is that aesthetic judgement rests, fundamentally, on a felt response to an object (Kant 1987: §1). Aesthetic judgement on an object cannot be passed second- hand: it requires personal acquaintance. The testimony of others may make me conﬁdent that I will ﬁnd a newly published novel rewarding; but I cannot declare it to be so until I have read it myself. The condition of felt response is reﬂected in the fact that an aesthetic judgement, like an avowal of emotion, can be made sincerely or insincerely. The second feature, closely connected to the ﬁrst, is that rules or principles play no role, or only a highly diminished role, in aesthetic judgement (ibid.: §8). It cannot be inferred that a piece of music is rapturous simply because it is in a particular key and orchestrated for certain instruments; or that a painting is dynamic because its compo- sition has a certain geometric form. No laws connect the aesthetic qualities of an object with its sensory, non-aesthetic properties. The point is not just that we are ignorant of rules for aesthetic judgement, or unable to agree on them, but that there would be little point in trying to formulate a set of rules: we could not use rules to produce in ourselves felt responses to objects; we could not, in cases of disagreement, reasonably ask another person to relinquish his or her judgement on the grounds that it conﬂicted with the rules; and whenever our own responses departed from the rules, we would quite rightly repudiate the rules rather than our responses. The fact that rules are otiose in determining the aesthetic qualities and value of objects reﬂects the fact that aesthetic interest is directed essen- tially towards particular objects, appreciated for their own sake, and not towards the formulation of general truths. To judge according to rules is to miss the uniqueness of objects, and so to fail to judge aesthetically. The place of rules, in guiding aesthetic judgement, is taken by particular, exemplary aesthetic objects; ‘established models’, as Hume calls them. Hume and Kant agree, further, in rejecting the view (of earlier, classical and ratio- nalist aesthetics) that aesthetic qualities are objective. Aesthetic objectivism maintains that aesthetic qualities are properties inhering in objects, and that aesthetic experience gives us knowledge of these properties. These properties may be identiﬁed with the object’s formal properties, such as ‘ideal proportion’; or they may be regarded as sui 232 AESTHETICS generis and irreducible to formal properties (Moore 1984: §§112–21; McDowell, in Schaper 1983). (These varieties of aesthetic objectivism bear comparison with natu- ralism and intuitionism in ethics, respectively.) Aesthetic subjectivism denies that aesthetic qualities inhere in objects, and maintains that what it is for an object to be beautiful is for it to yield a certain response in the subject. In aesthetic experience I am affected by the object, and my response does not consist in knowledge of its properties. The object’s formal and other non-aesthetic prop- erties are merely what occasions the response. Aesthetic subjectivism, we will see, comes in various degrees of sophistication (which again may be compared with differ- ent forms of subjectivism in ethics). Several very powerful considerations militate in favour of aesthetic subjectivism. Firstly, as Hume observes, and as we all recognize, the tastes and verdicts of different individuals and cultures vary enormously. (Lest this be doubted: Johnson said of Shake- speare that he has ‘faults sufﬁcient to obscure and overwhelm any other merit’; Voltaire compared Shakespeare’s works to a ‘dunghill’.) If aesthetic qualities are really out there, inhering in the object, why do we not all pick up on them? The point is not just	Blackwell
that divergence is a fact of aesthetic life, but that we lack agreed methods for bringing divergent judgements into line. As was said earlier, rules cannot be appealed to. And aesthetic disagreement is unlike disagreement about colour, where conditions of phys- iological normality and standard lighting can be checked independently. Aesthetic qualities are deeply elusive: the ‘positioning’ of the observer needed to discern aesthetic qualities involves a plethora of temperamental and culturally parochial factors, all strongly indicative of subjectivity. The second point was stressed by Kant (1987: §1). Pleasure is essential to the experience which grounds an aesthetic judgement. Pleasure, like pain, is, however, a mental state that does not represent a property of its object; as Kant says, pleasure ‘designates nothing whatsoever in the object’. The necessary connection of aesthetic judgement with pleasure is readily explained by the subjectivist, but it presents aesthetic objectivism with an enormous difﬁculty: how can there be properties inherent in objects which necessarily generate pleasure simply through being apprehended – if not on the theological supposition that the world has been metaphysically tailored to delight us? Such properties would be metaphysically queer in the extreme. (This objection to aesthetic objectivism may be compared with Hume’s objection to ethical objectivism, that it fails to capture the necessary connection of moral judgement with the will.) This objection can be ampliﬁed. Objectivism, because it holds that aesthetic qualities inhere in objects, necessarily introduces a distinction between reality and appearance, between how things are and how things seem. Objectivism therefore implies that, even when an aesthetic judgement has withstood all of the toughest tests, we can still be mistaken about it: it is still possible that the object lacks the aesthetic quality attributed to it and is really aesthetically worthless. But this seems mad. It is simply not intelligi- ble that our greatest art could ‘really’, appearances to the contrary, be worthless. Suc- cessful ‘appearances’ of aesthetic quality are as good as – indeed they are the same as! – the real thing: for aesthetic reality precisely consists in appearances. Put another way, the usual motivation for objectivism in philosophy – namely, showing that our experiences and theories put us in touch with the world as it really 233 SEBASTIAN GARDNER is, independently of our means of representing it – has no relevance to the context of aesthetic experience, which has instead a strong similarity to the context of personal relationships, in which rewards other than the gaining of knowledge are sought primarily. The cognitive demand that mind should ‘ﬁt the world’ does not apply to aesthetic interest, which is in this respect closer to desire than belief – what we want is for the world to ﬁt us. If, however, aesthetic subjectivism is common-sensical and philosophically plausible, it appears to carry a highly uncommon-sensical and unwelcome implication. If aes- thetic qualities do not inhere in objects, what is to prevent the same object being judged beautiful by one person and ugly by another – both being equally justiﬁed in their pro- nouncements? If, as Hume puts it, ‘All sentiment is right’, there is no such thing as right- ness or wrongness in aesthetic matters. In that case, an aesthetic judgement merely reports the speaker’s mental state, and aesthetic preferences are on a par with gusta- tory preferences. It follows that aesthetic judgements cannot be contested or supported with reasons. This position, like emotivism in ethics, to which it corresponds, is incon- sistent with common sense (Kant 1987: §56). Thus the problem of taste: of how to maintain the subjective felt basis of aesthetic judgement, without collapsing into unrestricted RELATIVISM (pp. 395–7). Hume and	Blackwell
Kant respond to the problem of taste in different ways. Hume contends that a ‘standard of taste’ – a basis for accepting some judgements of taste as correct and rejecting others as incorrect – is both philosophically defensible and visibly at work in the way that we actually conduct our aesthetic affairs. The stan- dard of taste lies not in the object but in the sensibility of the subject. Aesthetic sensi- bility varies in its ‘delicacy’: ‘When the organs are so ﬁne as to allow nothing to escape them, and at the same time so exact as to perceive every ingredient in the composition, this we call delicacy of taste’ (Hume 1965: 11). A correct aesthetic judgement is one that issues from a delicate sensibility operating under ideal conditions, which include the cultivation of taste through practice, the ability to make relevant comparisons, and freedom from prejudice. We defer to the judgement of those whose sensibilities we acknowledge to be superior, that of the ‘critics’. (The concept of an IDEAL OBSERVER (p. 470) plays an analogous role in Hume’s moral theory.) It follows that for Hume, aesthetic judgements do not identify aesthetic qualities inhering in objects, but nor do they simply report the subject’s experiences. An object’s possession of an aesthetic quality consists in its being ‘ﬁtted’ to generate a certain response in us. This distances Hume from the kind of unsophisticated subjectivism which ends up in relativism. Hume’s solution to the problem of taste depends on contingent facts of nature. Firstly, the fact that some ‘particular forms or qualities [in objects] are calculated to please, and others to displease’. Secondly, the contingent uniformity of human sensi- bility, the sameness in ‘the original structure of the internal fabric’ of our minds – by which is meant, not that we are all equally competent aesthetic judges, but that our sensibilities are all of a kind. Where sensibilities differ, the differences are not arbitrary: aesthetic judgements diverge because of differences in delicacy, and other determinable deﬁciencies of taste; if the sensibilities of two subjects are equally delicate and their conditions of judgement otherwise the same, their judgements will coincide. 234 AESTHETICS Kant and the Normative Aspect of Aesthetics Kant rejects the kind of account proposed by Hume. Recall that the problem of taste is one of justiﬁcation. The question is, on what grounds can we regard certain aesthetic judgements as being correct? For Kant, there is nothing in Hume’s position which accounts for this, the normative aspect of aesthetic judgement. Hume does not validate the claim, implicit in all aesthetic judgement, that one’s response is appropriate to the object, and that others ought to concur. Kant interprets this demand for agreement strongly, as applying to all other people without exception. On Kant’s view, Hume offers only a causal, psychological explanation of why we do, as a matter of fact, defer to the judgements of ‘critics’ – he does not say why we ought to do so. Kant’s solution to the normative problem is, in essence, very simple, although his pre- sentation of it is extremely intricate. Suppose that, in making an aesthetic judgement, we abstract from everything that might pertain to our contingent, natural, individually variable constitutions, and base our judgements solely on conditions that are strictly uni- versal, in the sense of being available and common to all human beings. Suppose, that is, we base our aesthetic judgement on the bare form of the object and its interaction with our basic, universal mental powers of perception and understanding. (Kant’s rea- soning here recapitulates his analysis of moral judgement in terms of the categorical imperative.) Kant argues that this ‘universal’ standpoint can be achieved through freeing our awareness of the object from desire and practical concern (what he calls ‘disinter-	Blackwell
estedness’), and from our conceptual understanding of it. When these stringent condi- tions are met, the judgement we make is valid for everyone. We then have what Kant calls a ‘pure judgement of taste’, which has ‘universal validity’. Kant argues that it is indeed possible for the mere form of an object to delight us: certain perceptual forms stimulate our mental faculties optimally, by engendering in us a ‘harmonious free play of imagination and understanding’, awareness of which is pleasurable. Kant also gives a metaphysical interpretation of aesthetic experience: it makes us conscious of a connection that we have with the world, and with one another, which lies beyond the empirical world. Beauty therefore has a semi-religious and moral signiﬁcance for Kant (he calls beauty the ‘symbol of the good’). Even if we grant that Kant’s theory of mental harmony accounts for the experience of beauty, a cost attaches to Kant’s solution to the problem of taste. Purifying aesthetic judgement in the way that Kant requires leaves us with an austere, if not impoverished view of aesthetic experience – as consisting of nothing but an estimation of form. This limitation shows up in Kant’s inability to do justice to the psychologically rich diversity of experience which art offers. We are left, then, with a choice: between accepting Hume’s account, without (Kant argues) any way of securing the normative aspect of aesthetic judgement; and accept- ing Kant’s account, at the price of excluding from aesthetic experience our cultural identities, and everything in our psychological constitutions that is not strictly univer- sal (which, the Humean will argue, leaves all too little). The task of providing a single account which incorporates and combines the insights of Hume and Kant remains. A number of other issues remain unresolved within the framework of aesthetic subjectivism, including the formulation of the doctrine itself. Granted that aesthetic 235 SEBASTIAN GARDNER qualities are subjective, quite how subjective are they? Are they, for instance, on a par with colours and other secondary qualities of objects? Some theories maintain that they are. Others deny that aesthetic qualities are on a par with colours and compare them with ‘looks’ or ‘aspects’ of objects, or describe them as ‘powers’ to produce experiences. Another, more radical, possibility is that aesthetic judgements are not descriptive at all. WITTGENSTEIN (chapter 39) (1978) suggests that aesthetic judgements are more like gestures and exclamations. On this view, the function of an aesthetic judgement is not to say anything about an object, but to ‘put across’ an experience. (These options are explored in Sibley 1959, 1965; Sibley and Tanner 1968; Wollheim 1980: essay 6; Hungerland, in Osborne 1972; Meager 1970; and Scruton 1974.) It is also necessary, as both Hume and Kant saw, to clarify the nature and role of reasons and criticism in aesthetic contexts. That aesthetic judgements can be justiﬁed distinguishes them from mere likings, and makes them, in a broad sense, rational. It is the job of the critic to identify the sources of a work’s aesthetic qualities and determine what responses are appropriate (a further role for criticism, to be discussed later, is inter- pretation). But aesthetic reason-giving differs fundamentally from reason-giving in other contexts and has a number of peculiarities, which make it hard to understand (Beardsley 1982: ch. 12; Scruton 1979: ch. 5). It does not consist in inferring one proposition from another; it makes no use of rules, unlike moral reason-giving; and it makes no use of induction or generalizations, unlike reason-giving in science. Even the notion of consistency has little application to aesthetic judgements: liking Wagner does not mean that one ought to dislike Mozart. The problem, in sum, is that aesthetic reasons operate independently from all of the conditions that appear integral to the	Blackwell
very concept of a reason. Two further unresolved issues may be pointed out in conclusion. The ﬁrst, bequeathed by Kant’s notion of disinterestedness, concerns the importance of the concept of an ‘aesthetic attitude’. Some theorists have claimed that aesthetic experi- ence consists in the adoption of a special attitude in which objects are attended to dis- passionately and ‘for their own sake’. Indeed, Schopenhauer identiﬁes the point of art with the metaphysical liberation from life that aesthetic contemplation, which does not involve the will, brings in its wake (Schopenhauer 1969, 1: third book). The distinctive phenomenology of aesthetic experience makes this idea alluring, but what can be said about the aesthetic attitude, other than that it differs from and excludes practical and cognitive attitudes? Saying that an object is enjoyed ‘for its own sake’ really means that it is ‘not enjoyed for any ulterior end’. Because descriptions of the aesthetic attitude tend to remain fundamentally negative, it is unclear what value the concept has, stripped of the metaphysical interpretation that makes it signiﬁcant for Schopenhauer (see, however, Beardsley 1982: ch. 16). A second issue concerns the concept of beauty, which has undergone a dramatic reversal of fortunes in the history of aesthetics. Classical aesthetics takes it for granted that beauty is the only, or at least the fundamental, aesthetic quality. Some modern writers propose by contrast that ‘beautiful’ is merely a catch-all term, roughly equiva- lent to ‘aesthetically commendable’, and that there is, as ordinary language implies, a limitless plurality of aesthetic qualities, encompassing elegance, grace, poignancy and so on. There is, however, a philosophical issue here which needs to be addressed. The idea that beauty is not just one aesthetic quality among many, but is somehow pre- 236 eminent and accounts for the unity of aesthetic experience, is compelling, and anyone who denies this in favour of the ‘pluralist’ view is challenged to ﬁnd an alternative account of the unity of the aesthetic. A recent defence of the concept of beauty is in Mothersill (1984). AESTHETICS 2 The Essence of Art HEGEL (chapter 33) asserts that art ‘pervades what is sensuous with mind’, and that art accordingly has ‘a higher rank than anything produced by nature, which has not sustained this passage through the mind’ (Hegel 1993: 15, 34). Whether or not Hegel’s view of the superiority of art is justiﬁed, the philosophy of art, to which we now turn, shares Hegel’s conception of art as a synthesis of mind with something other than itself, in which the mind ‘recognizes itself ’; and his rejection of the view that a work of art is simply an artefact possessing aesthetic qualities of a kind also found in nature. Theo- ries of art attempt to elucidate the nature of the connection with the mind which makes an object a work of art. Such attempts assume that the concept of art picks out more than a merely nominal, or historically accidental phenomenon – that art has, in short, an essence. 2.1 Deﬁnitions of art Attempts have been made to capture the essence of art in a simple deﬁnitional formula. Two representative examples are the deﬁnition of art as the imitation of beautiful nature, and the deﬁnition of art as the communication of feeling. If we consider such deﬁnitions, it is clear that they are wide open to counter-examples. Not every work of art is an imitation of beautiful nature, or communicates feeling: novels are rarely if ever beautiful, instrumental music is not imitative and much visual art does not communicate feeling. And even if these formulae provided necessary conditions for art, they could hardly provide sufﬁcient conditions: some industrial machinery is beautiful, waxworks are imitative, racist propaganda communicates feelings. One response to the failure of deﬁnitions of this simple kind is scepticism. It has been	Blackwell
claimed that the difﬁculty of deﬁning art is a reﬂection of there not being an essence of art. Proponents of this view (discussed in Tilghman 1984) have often appealed to Wittgenstein’s notion that family resemblances between instances of a concept may be all that there is to be found. A recent, much-discussed attempt to deﬁne art – which is built on the Wittgenstein- ian denial that art has an essence, but maintains that a deﬁnition is nevertheless possible – is the institutional theory. The institutional theory accepts that works of art cannot be deﬁned in terms of their intrinsic, perceptible properties, and turns instead to their extrinsic, social properties. The theory says that something is a work of art if a member of the artworld has conferred on it the status of being a candidate for appreciation, or if it has been created in order to be presented to the artworld (Dickie 1984). The artworld is a nebulous entity that includes artists, critics and some portion of the general public. ‘Work of art’ is, on this account, an ‘honoriﬁc’ term: being a work of art is a non-per- ceptible property which depends upon an object’s place in a cultural context (analogous 237 SEBASTIAN GARDNER to being ‘in public ownership’ or ‘sacred’). This approach receives an additional stimulus from the way in which some avant-garde art of this century – paradigmatically, Duchamp’s ready-made Fountain, a mass-produced urinal entered for exhibition unal- tered by the artist – appears to have repudiated all traditional criteria for art. There is, however, something missing at the heart of the institutional theory. It tells us that members of the artworld bestow the status of art on some objects rather than others; but not why they do so. It thus leaves out of account the reasons for calling something a work of art – the conditions taken to justify application of the concept. This is something that ought to ﬁgure in a philosophical, as opposed to a sociological, treatment of the concept of art (Wollheim 1980: essay 1). Proponents of the institutional theory seek to deﬂect this criticism by saying that they are concerned with the ‘classiﬁcatory’ sense of art, not the ‘evaluative’ sense in which to call something a work of art is to recommend it for appreciation. But this invites the further objection that it is simply a mistake to separate the classiﬁcatory sense of art from the evaluative. Evaluation is just as integral to the concept of art as it is to moral concepts. We do not ﬁrst classify objects as art, and then discover that they happen to be aesthetically rewarding: conceptually, there is only one move here. What moral is to be drawn from these attempts to deﬁne art? It cannot be ruled out that a complex deﬁnition, perhaps combining the concepts of imitation and communi- cation in a complex way and adding further conditions, could be devised to encompass all and only those objects that we consider to be works of art, and at the same time respect the evaluative nature of the concept of art. But what the difﬁculty of reading off an adequate deﬁnition of art from the surface of our ordinary conception of art may be taken to show, more importantly, is that no deﬁnition of art can be expected, and in any case cannot hold much interest, in advance of a theory of art. A theory of art does not require there to be a simple, manifest property shared by each and every work of art; it allows that what uniﬁes art – its essence – may not be visible at its surface, and probes beneath the surface to locate it. A theory of art may take one of two forms. It may aim directly at developing a single concept capturing the essence of art. Or it may proceed by building up from an exam- ination of the various speciﬁc dimensions of works of art. The ﬁrst, more traditional approach is explored below; the second, more modest approach favoured in contempo- rary aesthetics, is explored in the following section. 2.2 Theories of art	Blackwell
Of the numerous and enormously varied theories of art that have appeared in the history of aesthetics, three may be singled out as having greatest importance. These are the mimetic, formalist and expression theories. Even if the concept of imitation, as ordinarily understood, does not straightforwardly deﬁne art, it may be supposed that the concept can be developed beyond its ordinary scope. On this claim rests the theory that art consists in mimesis – a Greek term ren- dered approximately by ‘imitation’, ‘copying’ or ‘representation’ (see Plato 1955; Aristotle 1987; and chapter 23 of this volume). The object of mimesis is usually iden- tiﬁed with nature, inclusive of human nature. Although, as observed earlier, some art, such as instrumental music, appears to be non-representational, it is open to the 238 AESTHETICS mimetic theorist to contend otherwise: in antiquity it was thought that music imitates the order and harmony of the cosmos and the soul. The mimetic theory construes the connection of art with the mind asserted by Hegel in the following way: works of art carry over the mind’s fundamental function of representing the world. Granted that the concept of representation is open to being extended, the mimetic theory is open to an objection (Hegel 1993: LXI–LXVII). Even if we accept that it is natural to enjoy imitation and the skill exhibited therein, as Aristotle observes, why should we ﬁnd imitation valuable in the special way that art is found valuable? The values connected conceptually with representation are cognitive, truth-orientated values such as accuracy and comprehensiveness. These have some role in art – verisimilitude of plot and character in literature for instance – and certainly truth is something we value. But truth does not capture the real interest of art. A work of art must come into its own in the ﬁeld of our experience, and not disappear from our attention in the manner of a transparent window on to the world or vehicle for communicating truths about it. The mimetic theory thus appears to misconstrue the value of art. The mimeticist’s conception of representation accordingly needs to be tightened up. Representation which is artistic must be circumscribed in subject and mode: it must be of particular kinds of things, represented in a particular way. It is therefore no accident that mimetic theorists have characteristically gone on to claim that art must idealize its subject matter. But once the mimetic theory has undergone this modiﬁcation, the essence of art has been shifted away from bare representation, in the direction of whatever it is about art that enables it to idealize its subjects. A strong candidate for this role is artistic form. Where the mimetic theory ties art down to the real world, formalism allows the work of art to ﬂoat free, insisting that only form, the complex arrangement of parts unique to each individual work, has artistic signiﬁcance. Only what is internal to the work is relevant to its status as art; any outward references, to a real or imaginary world, are irrelevant. The concept of form can be made more or less narrow; at its narrowest, only sensory parts and their relations are intended. Formalists differ over the continuity of form in art with form in nature: Kant holds that artistic form must be recognizably of a kind with natural form; Bell (1914: ch. 1), the boldest recent exponent of formalism, holds that artistic form, labelled ‘signiﬁcant form’, is in effect exclusive to art. Appreciation of art consists, for the formalist, not in merely recognizing form, but in responding to it: Kant’s theory of mental harmony has already been mentioned; Bell posits a unique kind of ‘artistic emotion’ accompanying the recognition of signiﬁcant form. According to formalism, what the mind recognizes of itself in works of art is its power of locating order in perception, exercised to a specially heightened degree.	Blackwell
The weakness of formalism most frequently indicated by its critics concerns the concept of form, which is argued to be too indeﬁnite to play the role that is asked of it. Asked to specify the kind of form that matters in art, the formalist uses notions such as ‘balance’ or ‘uniformity amidst variety’; but it proves impossible to deﬁne these in a way that prevents them from applying to any object whatsoever. The formalist is conse- quently pushed to declare that artistic form is indeﬁnable. Bell embraces this claim, but his theory is then charged with vacuousness: he explains signiﬁcant form in terms of the ‘artistic emotion’ that it induces, but then refers us back to ‘signiﬁcant form’ for our understanding of artistic emotion. 239 SEBASTIAN GARDNER Whatever the force of this point against Bell, there is another objection to be pressed. Is our interest in form really as uncontaminated with ulterior, worldly concerns as for- malists suppose? Formal values may sometimes be of self-sufﬁcient aesthetic interest, but much more commonly they serve non-formal ends: form is the vehicle through which a work articulates its non-formal meaning, apart from which form tends to become artis- tically uninteresting and merely decorative. In other terms, the attempt to disentangle form from content leaves, on the side of form, an insufﬁciently signiﬁcant residue. This objection to formalism may be pressed on behalf of the mimetic theory; but it also points in the direction of the expression theory. The connection of art with feeling, we said earlier, cannot consist in a straightforward equation of art with the communi- cation of feeling. The expression theory of art seeks to offer a more sophisticated and persuasive account of the central place of emotion in art. The rudiments of the expression theory lie in writings of the Romantics, but it was ﬁrst formulated philosophically by Croce (1992), who bound it up with the tenets of German idealism. The more accessible version of the theory presented by Collingwood (1937) detaches it from METAPHYSICS (chapter 2), and understands artistic expression as a special form of self-expression. Artistic expression is a process in which the artist begins with an indeﬁnite and inchoate emotional state, for which he or she labours to ﬁnd a uniquely apt concrete articulation, and in so doing transforms his or her mental state into something deﬁnite, tangible and intelligible. What the artist creates does not describe his or her state of mind, so much as incorporate it; analogously to the way in which bodily expressions such as smiles and grimaces embody mental life. Since the product of expression cannot be known before the process is complete, expression cannot consist in exercising a technique: it must take whatever particular shape is commanded by the particular emotion which the artist’s mind is impelled to clarify. Because expression is not undertaken with any further end in view, artistic cre- ation contrasts with instrumental activities, in which means and ends are distinct; these Collingwood calls ‘craft’, and opposes to ‘art proper’. Mistaken, ‘technical’ con- ceptions of art – which include the conception of art as mimesis or communication of feeling – arise from a failure to grasp this distinction. On the side of artistic appreciation, the theory claims that the audience retraces in the course of appreciating a work of art the route pursued by the artist: their apprecia- tion re-enacts the artist’s creative process and thereby retrieves his or her psychologi- cal state (Elliott, in Osborne 1972). The relation of the audience to the work of art thus mirrors the artist’s understanding of his or her own mind. The capacity of a work of art to transmit the artist’s psychological state is a necessary consequence of successful expression. The expression theory, by giving primacy to the perspective of the artist rather than (as on the mimetic and formalist theories) that of the audience, offers a	Blackwell
very strong interpretation of Hegel’s claim that the mind ‘recognizes itself ’ in works of art: works of art do not merely exhibit mental features, they, as it were, contain mind. The expression theory’s emphasis on the psychology of the artist exposes it to criti- cism. The mimeticist will object that not only emotions, but also ideas, which refer to the world, are expressed by art, whose legitimate subject is not restricted to the artist’s own mind. At the other extreme, the formalist will challenge the expression theory to say how a supremely self-contained and self-sustaining work such as a Ming vase, can be construed as a product of personal expression. 240 AESTHETICS The ﬁrst objection can be met by explaining that what the theory means by emotion is any mental state, however permeated with concepts and thought, that has some emo- tional charge. The condition that emotion be present is plausible, for most people would agree that mere beliefs without any psychological resonance do not provide sufﬁcient material for art. The scope of the thoughts embedded in the artist’s emotion is fur- thermore left open, which means that the artist’s moral and spiritual view of the world and conception of life are proper material for expression, and that what a work of art expresses need not include a personal, biographical reference to the artist. The formalist’s objection cannot be met by expanding the terms of the theory, and obliges the expression theorist to adopt a semi-stipulative measure. The theory must declare that objects possessing only formal virtues do not qualify as full-blooded instances of works of art. This move is obviously acceptable only on the assumption that art’s centre of gravity lies in the psychology of the artist; it will be rejected by the formalist as begging the question, and as betraying the theory’s fundamentally parochial, pro-Romantic bias. Of the three theories considered, it is fair to say that – if a monopoly is to be granted – then the expression theory has strengths that give it the edge over the mimetic theory and formalism. These strengths also account for the theory’s continued hold on our thinking about art. It may also be pointed out in its favour, that the expression theory allows the understanding of art to draw on the resources of psychological theory (Wollheim 1987); an ultimate verdict on the theory will depend in part on what value is found in such developments. The Semiotic Theory of Art A fourth theory, which is not traditional but dominates much contemporary discourse about art, should be mentioned in conclusion. This is the semiotic theory of art, which proposes that works and forms of art be analysed in terms of logico-linguistic categories such as signiﬁcation, reference, denotation, and syntactic and semantic rules. Goodman (1976), the principal exponent of the semiotic theory, analyses representation and expression in such terms. The semiotic theory of art is also assumed by structuralist and poststructuralist literary theory, which grafts linguistic science and philosophical theories of language onto the study of literature. On the semiotic theory, art is a symbol system much like language, and not just in the metaphorical sense in which an expres- sion theorist might grant that art is a ‘language of emotion’. What the worth of the semiotic theory turns on can be indicated with reasonable assurance. The theory derives its motivation from, ﬁrstly, a rejection of the primacy of psychological and experiential concepts (these, it claims, are not autonomous and need to be understood in terms of language and signiﬁcation); and, secondly, a belief in the radically conventional nature of art (which it grounds on a blanket rejection of philo- sophical realism). Both strands are explicit in the work of Goodman, and in structural- ism and poststructuralism. It follows that the semiotic theory’s austere perspective on art	Blackwell
must be rejected by anyone who wishes to maintain that art is essentially connected with certain forms of experience; as it may also be, on the supposition that human psychol- ogy sets universal, non-conventional parameters to art. This last idea provides the basis for a ﬁfth theory of art, the naturalistic, which will be described in section 4 below. 241 SEBASTIAN GARDNER 3 The Dimensions of Art Recent work in analytical philosophy of art has adopted the second, more piecemeal and empirical, approach described earlier to the construction of a theory of art. Rather than aim directly at a general, overarching characterization of art, it has concentrated on the speciﬁc dimensions which constitute works of art, in the belief that an accurate its parts. picture of art as a whole will emerge out of a proper understanding of Representation, expression and meaning have been closely analysed, in terms of the speciﬁc forms that they take in each of the arts. This section traces the debates sur- rounding these concepts in terms of the form of art most strongly associated with each: representation in painting, expression in music, and meaning in literature. (Issues concerning the ontology of works of art, the nature of ﬁction, emotional response to ﬁction, and metaphor, also belong in this context, and would be included in a fuller account of analytic philosophy of art.) 3.1 Representation When we look at Titian’s The Rape of Europa we do not see, or do not just see, pigment distributed across the two-dimensional surface of a canvas: we see a woman borne across the sea aloft a white bull. Paintings represent or depict things, both real and imaginary. We are concerned here with what paintings represent in a visual, as opposed to interpretative, sense. A painting of a man clutching a stone may represent St Jerome, and a lute with a broken string in a still-life may represent Discord. In order to make such iconographic identiﬁcations, however, a human ﬁgure and a musical instrument must ﬁrst be seen in the picture, and this is a perceptual matter; it is something that art historical scholarship could not do for us. Visual representation is something we take for granted. When we cease to do so, a philosophical question arises: what makes pictorial representation possible? The mimetic theory of art, it should be noted, does not answer this question: it merely assumes that pictorial representation is possible. The philosophical puzzle surrounding pictorial representation comes into focus when one reﬂects on the difﬁculties associated with the common-sensical view of how pictures work. Common sense says that pictorial representation consists in resemblance: a painting represents X by looking like X. This may at ﬁrst seem to be not just self-evident, but the only possible answer. This common-sense view of pictorial repre- sentation is reﬂected in the idea that deception provides the test for successful repre- sentation – as illustrated by the story of Zeuxis’s painting of vines, which, according to Pliny, deceived the birds into ﬂying down to consume the grapes he had depicted. It is, however, simply and plainly false that canvases marked with pigments resem- ble the things that they represent. The Rape of Europa does not even begin to share the physical dimensions of the scene that it represents! In fact, each canvas resembles nothing so much as other canvases – but it does not, of course, represent them. This objection may seem too blunt, for it may seem to have missed the obvious point that for a canvas to represent something, it must be seen, not as a mere physical object, 242 AESTHETICS but as a picture. But this just restates the problem, which now becomes: what is it for a canvas to be seen as a picture? It may then be ventured that what the resemblance theory really means, is that a painting of X represents X when it affords us an experience which is like the experience	Blackwell
of really perceiving X. Sometimes this is put by saying that the canvas gives us an ‘image’ which is the same as that which we would receive from the real thing. Relocating the resemblance at the level of an experience or image, rather than that of the canvas, does not, however, advance our understanding for, once again, what is wanted is an account of how the two-dimensional, differentiated surface of Titian’s painting can be made to ‘have the look’ – afford the visual experience or create the image – of a woman carried on the back of a bull. Very quickly it comes to seem that resemblance does not provide the explanation of pictorial representation. In its place, theorists turn to other, less obvious notions. (However, for a recent defence of pictorial representation in terms of experienced resemblance, see Peacocke 1987.) The case against resemblance was ﬁrst set out, at length and with sophistication, by Gombrich (1960). Gombrich attacked in particular the assumption, which accompa- nies the resemblance view, that there is such a thing as an ‘innocent eye’ – that an artist can simply ‘copy what he sees’ by attending to his visual experience, in advance of any interpretation of the world. Goodman (1976: ch. 1) carries this line further by rejecting altogether the idea that pictorial representation is a matter of perception. On Goodman’s account, depiction is a species of denotation: knowing what a picture represents is purely a matter of inter- pretation, which requires a grasp of a ‘symbol system’. Realism in painting is just a function of the familiarity of a symbol system, the ease with which it imparts information. As was said earlier, the motivation for Goodman’s semiotic view is very general. There is, however, much to query in his claim that pictures are interpreted. It seems to us that pictures give us visual experiences, and that this is how they differ from prose descriptions, maps, company logos, road signs and other visual symbols that need to be read. Goodman’s theory implies that this impression of difference is an illusion. But if so, what does the illusion of having a visual experience consist in – if not in having a visual experience? Gombrich himself does not assimilate pictorial representation to interpretation com- pletely. On Gombrich’s account, a painting represents X by giving us an illusion of X. How illusions are created is explained in terms of the representational practices evolved in the history of art. Art develops through a process of ‘making and matching’. Visual ‘schema’ are originally fabricated semi-arbitrarily in order to represent objects, without resemblance playing any kind of role. They are then modiﬁed by artists in the light of their perceptual experience, a historical process which results in increasingly life-like representations (a process which Gombrich models on POPPER’S (pp. 287–90) concep- tion of the development of scientiﬁc theories). Resemblance lies, if anywhere, at the end of the history of pictorial art, not at its beginning. It is true that some visual representations – such as Escher’s trompe l’oeil drawings – create illusions, but it is not true that most paintings have this effect: in looking at a painting we do not ﬁnd ourselves having to correct a tendency to mistake its content for 243 SEBASTIAN GARDNER reality, as the concept of illusion would imply. Furthermore, the illusion theory entails that in seeing what a painting represents, we stop being aware of the material canvas, as the illusion takes hold. This, however, contradicts the fact that appreciating a paint- ing involves simultaneous, complex awareness of its brushwork and what it represents (see Wollheim 1974: ch. 13). Representation and Seeing In The weaknesses of the views considered so far suggest that we need to posit a form of visual perception which is, as it were, inherently imaginative. Wollheim’s theory does this (Wollheim 1980: essay 5; 1987: 46–77). It proposes that pictorial representation	Blackwell
engages a species of perception which Wollheim calls ‘seeing in’. Wollheim argues for this claim as follows. An important clue as to the underlying nature of the experience of painting lies in the fact that paintings give us experiences of something absent or non- existent. Now this is something that occurs also in dreams, daydreams and hallucina- tion. It may be ventured that pictorial representation exploits and cultivates a power that the mind possesses innately: the power to generate visual experiences out of itself. When this capacity is exercised in the course of perceiving the external world and fused with perception of external objects – as it is in numerous contexts outside painting: things are seen in Rorschach ink-blots, clouds and damp-stained walls – we then have seeing in. Seeing in is not involved in reading maps or interpreting visual signs; here there is no experience of an absent object. On this theory, what it is for Vermeer’s View of Delft to represent a townscape is for the canvas to be marked in a way that allows us to see a townscape in it. Seeing in does not presuppose a resemblance between the canvas and what the painting represents; according to Wollheim, no systematic account can be given of how the marking of a canvas determines what is seen in it. Wollheim (1987) proceeds to develop a general account of painting which shows how its value as an art presup- poses its rootedness in seeing in and other psychological powers. 3.2 Expression It is not long before the fugue recedes into its initial calm. Soon, a ﬁnal crescendo leads us to what appears to be a triumphant conclusion. But no; at the very moment when victory seems to be in our grasp, the music loses all conﬁdence, comes to a halt on a chord that is far from ﬁnal, and then sinks despondently back into the minor key. Could disillusionment be more graphically expressed? It is the cry of despair; the hoped-for consolation has failed to materialize. (Hopkins 1982: 383) This passage describes emotions expressed in the ﬁnal movement of Beethoven’s Piano Sonata Op. 110. The expressive qualities of works of art are a sub-class of their aesthetic qualities, distinguished by the employment of terms whose primary use lies in describing people’s emotions. Expressive qualities are integral to the beauty and meaning of music. Since the nineteenth century, many have believed that music’s incomparable power of expression secures for it a position of supremacy among the arts (Walter Pater claimed that ‘all art constantly aspires towards the condition of music’). 244 AESTHETICS Describing music in expressive terms exempliﬁes a broad tendency to talk about works of art as if they were sentient entities: they are described as having ‘organic unity’, ‘vitality’ and so on. And yet we do not, of course, really believe or mean to imply that works of art have feelings. So, given that a musical work is at one level just a sequence of sounds, just as a painting is nothing but a pigmented canvas, the funda- mental problem of expression in art, brought out most sharply by music, is the follow- ing: how is it possible for a work of art to express emotion? The obvious answer would seem to be that musical works have expressive qualities because they serve, as the expression theory of art says, as vehicles for expressing the emotions of the composer. This is, however, no answer. Even if the expression theory of art is accepted, nothing in it explains how it is possible for a composer to use patterns of sound to express emotion. It merely takes that fact for granted. If it does not help to think of the emotion expressed by music as located in the mind of the composer, it is equally mistaken to attempt to locate it in the mind of the listener: that is, to analyse a musical work’s expression of an emotion as its power to arouse that emotion in the listener. Bluntly: we are not ourselves made melancholic, or	Blackwell
frozen with grief, by music that we would describe in those terms – fortunately. The lis- tener does not have the emotion which the music expresses, in the same sense as she has emotions in life. In listening to music, the emotion which is expressed seems to be located ‘out there’ in the music, not in oneself. The heart of the problem lies in under- standing how emotion can be ‘objectiﬁed’ in this fashion – how it can be ascribed neither to the composer nor to the listener, but rather somehow hover, suspended, between them. Further reﬂection shows that being moved by music consists, typically, not in simply recognizing or sharing the emotion expressed by music, but in reacting to it – in the way that a person may respond to the emotional condition of another, for example, by responding to their distress with pity. The relation is one of sympathy rather than empathy. We are familiar with this structure in the context of representational art, where, in drama, literature and painting, we feel for the protagonist. Its appearance in music reinforces the difﬁculty of musical expression, since instrumental music precisely lacks the representational content that would make emotional response to music intelligible as a case of participating in a ﬁction (it has no characters or plot). At this point, one option is to reconceive the problem of musical expression as merely one of language: of ﬁnding an adequate logical characterization of how terms are used in the ascription of expressive qualities. An account of this sort – which appears to cut rather than untie the Gordian knot – is again advanced by Goodman (1976: 85–95). An alternative way out of the problem is musical formalism, a position defended by Hanslick (1986). Hanslick grants that music has emotional effects on listeners; but he denies that there is any genuine sense in which music expresses emotion. Hanslick has both an argument for denying that music can express emotion, and an account of why we should come to suppose falsely that it can. For music to express emotion, there would have to be an intrinsic relation between them. Hanslick argues that this is impossible, on the grounds that emotions are intentional: they have objects and involve thoughts and concepts; whereas instrumental music lacks objects and does not involve thoughts or concepts. Since music is incapable of presenting any object – what it presents is only itself – it cannot have an intrinsic relation to emotion. 245 SEBASTIAN GARDNER What music can do, Hanslick argues, is express what he calls ‘musical ideas’, which are its dynamic properties: ‘The content of music is tonally moving forms’ (ibid.: 29). Tonality provides a sort of auditory space within which the melody, harmony and rhythm of music move (see Scruton 1983: ch. 8; 1997: chs 2–3). Hanslick draws an analogy with the visual movement found in arabesque ornamentation. Like music, episodes of emotion have dynamic properties, in addition to and inde- pendently from their conceptual content, since they involve changes in intensity and bodily feeling. Because music and episodes of emotions share dynamic properties, it is possible for us to ﬁll in, arbitrarily, the musical idea expressed by a piece of music with some particular emotion supplied by our own emotional repertoire: giving us the illu- sion that the music expresses that particular emotion. For Hanslick, musical apprecia- tion stops properly at the beauty of musical form. Anything else is merely personal emotional association. Despite its subtlety, Hanslick’s repudiation of emotion is not easy to accept: describ- ing musical works in terms of their expressive qualities seems to be, not just natural, but indispensable for justifying our estimates of their artistic value. The thought that music is intrinsically expressive should not be given up unless absolutely necessary. Can anything be salvaged? Although music does not speak or behave, an analogy between the recognition of	Blackwell
emotion in music and in people can be suggested. Perhaps music derives its expressive power from its resemblance to emotionally expressive features of human bodily behav- iour, physiognomy and speech (Kivy 1989). That would explain the phenomenological similarity between hearing emotion in music, and perceiving emotion in the body or voice of another person: in both cases, mental life seems to have been made sensuously palpable. This theory is in a position to deﬂect Hanslick’s objection that music lacks the conceptual components necessary for an intrinsic connection with emotion. Bodily and vocal features can express emotional states without communicating their objects: we can recognize that a person is distressed from the way they look or sound, without knowing what they are distressed about. The human analogy theory soon encounters difﬁculties, however. It shares the weakness of the resemblance theory of pictorial representation – namely, the difﬁculty of locating relevant resemblances. What exactly are the features shared by music and human bodies? Just as canvases are unlike the things represented in pictures, a plain description of music in terms of key, rhythm and so on reveals little that is common to music and human bodies. It is true that both tempo and bodily movement can be slow, but such properties are no more emotionally deﬁnite than Hanslick’s ‘musical ideas’: if a sad piece of music is slow, what makes its slowness that of sadness rather than seren- ity? The objection then arises that signiﬁcant likenesses between music and human bodies can only be identiﬁed after the music has been redescribed in expressive terms, such as ‘ponderously slow’. Only once music has been experienced as expressive does it become possible to model it imaginatively on the human ﬁgure. Musical expression remains unexplained. A second theory that appeals to a resemblance between music and emotion is that of Langer (1942: ch. 8). On Langer’s theory, musical works are symbols of a special kind. Unlike the discursive symbolism of language, music shares the ‘logical form’ of 246 AESTHETICS emotion and, by virtue of doing so, articulates and presents emotion, as language cannot. Here it is the inner aspect of emotion, rather than its outer bodily expression, which music is said to mirror (‘Music is our myth of the inner life’: ibid.: 245). A similar problem, however, confronts Langer’s theory. In what non-metaphorical sense does emotion have ‘logical form’? And in any case, the notion of formal similar- ity is so plastic and open-ended that it can surely be found between any emotion and any piece of music. Again, the alleged isomorphism of music with emotion seems to be either non-existent, or insufﬁciently deﬁnite to explain musical expression. If neither the outer nor the inner aspect of emotion has a connection with music which accounts for its expressiveness, it may seem that enquiry grinds to a halt. Budd (1985) concludes, after intensive examination of the above theories, and others, that none succeeds). One remaining avenue, however, is to suppose that the connection of music with emotion is more direct than we have hitherto assumed it must be: that it does not take a detour via resemblance. A clue to the explanation of expressive qualities may lie in the fact that the mind has, as Hume put it, a natural propensity to spread itself on to the world, and experi- ence objects as emotionally coloured. Projection, as this may be called, is a well-attested psychological phenomenon. If this is joined with a second, equally Humean assump- tion, to the effect that certain objects in the world are ﬁtted by nature to serve as recipients for the projection of speciﬁc emotional states, then we are on the way to a conception of artistic expression as reposing on, and cultivating, a natural tendency of ours to dye objects in the world with our emotions (Wollheim 1980: sections 15–19; 1987: 80–7). If Hanslick’s repudiation of emotion is to be avoided, an assumption	Blackwell
along these lines seems to be at least necessary. 3.3 Meaning That works of art have meaning – over and above their representational content, and aesthetic and expressive qualities – is implied by the fact that works of art allow for, and call for, understanding. Questions of meaning in art are best approached through the concept of INTERPRETATION (pp. 384–90). Interpretation aims to identify the meanings of works of art, and is one of the functions of criticism. Questions of interpretation arise of course in all the arts – is Beethoven’s Missa Solemnis a Christian or a Roman- tic work? Does Titian’s The Flaying of Marsyus celebrate the triumph of spirit over body, or that of body over spirit? – but it is literature, and literary criticism, that put the issues surrounding interpretation in sharpest focus. Of all the arts, literature allows least scope for aesthetic response to operate inde- pendently of interpretation: a liking for a literary work not accompanied by so much as a rudimentary grasp of its meaning fails altogether to qualify as an appreciation of it. This is due to the fact that the medium of literature, natural language, is, unlike pigment, sound or stone, inherently meaningful; which, plausibly, allows literary works to bear more conceptually complex meanings than works in other media; for which reason literary works generate greater interpretative controversy. It is hard to ﬁnd a great literary work for which different interpretations have not been proposed. Is Hamlet correctly understood in terms of Freud’s theory of the Oedipus complex? Does Paradise Lost represent Satan as a moral being ‘far superior to 247 SEBASTIAN GARDNER his God’, as Shelley claims? Do Shakespeare’s sonnets comprise a uniﬁed narrative charting the course of Shakespeare’s own experience? Are Kafka’s writings religious or political parables, or articulations of the personal angst evident in his diaries and letters? Does a proper appreciation of Yeats’s poetry presuppose knowledge of his mytho-logical system? To each of these questions critics have returned opposite answers. Two questions are forced on us by differences in interpretation. Firstly, can bio- graphical information concerning the author serve as legitimate evidence for or against a certain interpretation? Secondly, can we talk of one interpretation as being the correct interpretation of a work? The starting-point for the ﬁrst question is the attack made by Wimsatt and Beards- ley on what they call the ‘intentional fallacy’ (in Newton-de Molina 1984). On their view, critics err in supposing that information about the author, or any kind of evidence ‘external’ to a poetic text (for instance, about its original cultural circumstances), can elucidate its meaning as a literary object; all that is pertinent to the meaning of a poem can be gleaned by careful attention to the words on the page. To suppose otherwise is to commit the fallacy of confusing a psychological fact about the author with a fact about the poem, of failing to distinguish a causal explanation of a work from a literary interpretation of it. Two opposing views of the relation between the meaning of a literary work (textual meaning) and the intentions of the author (authorial meaning) can be staked out. Intentionalism identiﬁes textual meaning with authorial meaning: what the text means is what the author meant (Wollheim 1980: essay 4). Anti-intentionalism denies that textual meaning is authorial meaning, and asserts its autonomy: textual meaning resides objectively in the work, ‘embodied in the language’, and has nothing to do, con- ceptually, with what the author may have meant (if the two happen to be the same, that is a coincidence without signiﬁcance, for the anti-intentionalist). These positions have connections with traditional theories of art: the expression theory is straightfor- wardly committed to intentionalism, formalism to anti-intentionalism (the mimetic	Blackwell
theory is uncommitted). What, then, are the main considerations in the argument over intentionalism? There is much to be said, but some of the most important points are the following. On the one hand, it seems right that a literary work, considered as a work of art, should be judged in terms of the experience of reading it; if meanings do not show up in an optimally sensitive reading of the text, then, although the author may have wished to convey them, they simply do not form part of what the text means. A poem stands or falls on its own merits. This is Wimsatt and Beardsley’s strongest argument for anti-intentionalism. It can, however, be countered by the intentionalist. While it is true that the author’s intentions must be realized in the text and concretely apprehended there in order for the work to be artistically successful, when a work is successful, intentionalism is true: what we grasp is the author’s meaning. The intentionalist may add that, although ‘external’ evidence cannot transform an artistic failure into a success, attention to such evidence may make it easier to see what meaning is contained in the words on the page. The intentionalist can also point out that, since we rarely approach literary works without some awareness of the author’s historical circumstances, other writings and 248 AESTHETICS so on, there is in fact no ﬁrm or deep distinction between internal and external evi- dence, contrary to what the anti-intentionalist supposes. The intentionalist’s case can be strengthened further by drawing attention to the fact that estimates of sincerity, maturity and perceptiveness are important for our responses to literary works. Literature is ﬂawed if it exhibits mawkishness, self-indulgence or crudity of moral vision; the ﬁne consciousness of life and ethical complexity of George Eliot and Henry James is valued. Literary vices and virtues are qualities of moral personality apparently ascribed in full-blooded intentionalist spirit to the author, in so far as he or she is manifest in his or her work. (Amplifying this approach, see Leavis 1986: part 4.) The anti-intentionalist will complain that here the intentionalist has once again con- verted an aesthetic into a non-aesthetic matter, and is treating the text as material for literary biography, as if it were the critic’s job to pass moral judgement on the author. We do value qualities such as sincerity in literature, but these are properties of texts: they indicate strengths in the work itself. We have, then, intuitions favouring each of intentionalism and anti-intentionalism. Neither set of arguments is obviously decisive. Wimsatt and Beardsley did not, there- fore, establish that intentionalism involves a fallacy. They did, however, deﬁne anti- intentionalism as a theoretical option, and indicate the methodological repercussions of the issue. If intentionalism is true, then the study of literature may cast its net wide and draw, in addition to literary biography, on other disciplines – history, psychology or anthropology – concerned with factors contributing, consciously or unconsciously, to the author’s meaning. If anti-intentionalism is true, then external information ought to be bracketed out. The second question raised by differences of interpretation, which concerns the notion of correctness, can again be formulated in terms of a sharp opposition. On the monistic view, there is for each work a uniquely correct interpretation. Whether it can be identiﬁed conclusively, and gain universal acceptance, is another matter. Monists say that where interpretations differ, and yet seem equally well-supported, their conﬂict can be overcome by a ‘super-interpretation’, which incorporates what is true in each of the interpretations which it supersedes. Pluralism denies that there is, even in principle, a uniquely correct interpretation for each work; it maintains that a number of different interpretations can meet the con-	Blackwell
dition of, not correctness, if correctness is taken to imply uniqueness, but ‘legitimacy’. Pluralists differ over the criteria for legitimacy of interpretation and how restrictive these should be. The arguments between monism and pluralism engage closely with the issue of intentionalism versus anti-intentionalism. The pluralist focuses in the ﬁrst instance on the notion of textual meaning. Cor- rectness of interpretation presupposes that textual meaning is determinate. Wimsatt and Beardsley help themselves to the assumption that literary works have objective, determinate meanings, but the pluralist challenges the monist to say where this deter- minacy comes from: for even if the meaning of each individual word composing a text is determinate – a claim which is itself vulnerable in view of the ubiquity of metaphor in literature – the meaning of the work as a whole is far from being a straightforward function of the dictionary meanings of the words composing it. As we would ordinar- ily put it, there is ‘room for interpretation’, and this, according to the pluralist, means 249 SEBASTIAN GARDNER that there is room for a number of interpretations. Easiest, then, is for the monist to embrace intentionalism: since, if intentionalism is true, a source of determinacy is readily supplied by the author’s intention. The pluralist’s second argument draws on a tenet of aesthetic subjectivism: namely, that felt response is at the core of the aesthetic. Literary interpretation should serve the kind of interest appropriate to literary works, and literary interest, as a case of aesthetic interest, centres on response. The proper goal of interpretation is therefore to enhance our experience of the work, to offer a reading which makes it mean as much as pos- sible to us. Since the perspectives of readers will always differ, the interpretation that is optimal for one reader will not be optimal for another. Multiple interpretations of a work are therefore to be expected, and should be welcomed. To meet this argument on its own terms, the monist would have to defend the – evidently controversial – claim that there is for each text one interpretation which is universally optimal. Arguments between monism and pluralism tend to converge on the following sce- nario. The monist says that if pluralism is true, then ‘anything goes’ in interpretation, an implication which the monist claims reduces pluralism to absurdity – The Ancient Mariner cannot be interpreted as a poem about issues of power and gender in late twentieth- century culture. The pluralist objects that the notion of a uniquely correct interpretation fosters dogmatism, and disregards both the richness of texts and the diversity of human interests which may legitimately be expressed in literary interpretation. These (unresolved) philosophical issues gain a special urgency from their relevance to the present, turbulent and logomachic climate of literary studies, where theoretical and ideological commitments are explicitly adopted and appealed to in interpreting texts. The dominant deconstructionist conception of literary meaning at work here is fundamen- tally a form of pluralism, adopted on wholly general philosophical grounds: the very concept of determinacy of meaning is rejected, in all, not just literary contexts. To this is added a radically idealist view of literary meaning which reverses the ordinary concep- tion of the relation of meaning and interpretation: the meaning of a literary work is conceived as created, rather than grasped by interpretation. The point of literature, on this picture, is not aesthetic: literature exists for the sake of interpretation. 4 The Point of Art We come ﬁnally to the question of the point of art. What value does art have? What is art for? This is not a demand for a ‘justiﬁcation’ of art in the same sense as a justiﬁca- tion of morality is thought to be required – that would be an inappropriate demand, since the creation and appreciation of art do not share the necessity of moral obliga-	Blackwell
tion. The question is rather whether we can articulate and vindicate our sense of the importance of art. The fact, observed earlier, that art is an evaluative concept, means that an account of the point of art is not a mere coda to aesthetic enquiry: it should fall out conceptually from a theory of art. That art does have some point is not beyond all doubt (Passmore 1991). It has already been seen how the mimetic theory encounters difﬁculties on this score. Posi- tive reasons for scepticism about the value of art are not hard to unearth: they derive from art’s essential connections with pleasure, play and imagination, its freedom from 250 AESTHETICS reason and practical purposes. The sceptic will insist that art has the same sort of value as any other form of entertainment. Plato’s notorious critique of art, in book ten of the Republic, goes further, by suggesting that art’s preoccupation with appearance entrenches our ignorance of reality, and that its effects may be psychologically and morally harmful (see Janaway 1995). The distinction of ends and means ﬁgures prominently in many discussions of the point of art, and is usually employed critically: some accounts are said to reduce art to a ‘mere’ means, others to recognize correctly that art is an end. This contrast is, however, not altogether felicitous, for two reasons. Firstly, art, so long as it has some value, can always be redescribed as a means to whatever value it realizes. Secondly, it is not clear what can be meant by describing art, or its apprecia- tion, as an end. Certainly individual works of art have to be approached in their own terms and contemplated ‘for their own sake’; but it would be a mistake to move from this fact about aesthetic attention, that it terminates in its object and does not ‘think ahead’, to the claim that art is its own point and cannot have extrinsic value. Formal- ists such as Bell and Hanslick, who describe works of art as ends in themselves, seem to make this mistake. What the denial that art is ‘merely a means’ – and the aestheticist slogan ‘art for art’s sake’ – may be trying to say, is that the value of art realized by art cannot be real- ized by anything else, or cannot be realized in the same way. Thus interpreted, the doctrine meets with our agreement: we recoil from hedonistic, moral, didactic or other instrumental attitudes to art, to the extent that these suggest that other things could be substituted for art without loss, or that the complexity of art is redundant. Tolstoy’s (1930) theory of art as the transmission of moral feeling, for example, makes little effort to distinguish art from other, potentially more efﬁcient, ways of achieving that end. It follows that so long as art is viewed as having a necessary role in relation to the kind of value in question, and the complexity of art is taken account of, there is nothing nec- essarily wrong with assigning a hedonistic, moral, didactic or other goal to art. Schiller (1989), for instance, argues that the goal of art, as a component of ‘aesthetic educa- tion’, is to overcome the metaphysical contradictions in human nature in order that we may attain full ‘humanity’; this goal is ‘extra-aesthetic’, but Schiller regards aesthetic education as the only way in which it can be achieved. What forms may accounts of the value of art then take? They may, ﬁrst, be divided according to how closely they relate the value of art to the values of life. In a perspective such as that of Leavis, art is properly and inextri- cably bound up with the values of life: the ‘raison d’être of the work’ is to ‘have its due effect and play its part in life’; the ‘essential business of criticism’ is to locate ‘the creative centre where we have the growth towards the future of the ﬁnest life’ (Leavis 1986: 283). At the other extreme, Bell holds that art is autonomous and has value only to the extent that it distances us from life: ‘to appreciate a work of art we need bring	Blackwell
with us nothing from life, no knowledge of its ideas and affairs, no familiarity with its emotions . . . In this world the emotions of life ﬁnd no place. It is a world with emotions of its own’ (Bell 1914: 25–7). The deeper distinction to be made among accounts of the value of art, however, is between those that are naturalistic, and those that are metaphysical; a distinction that was foreshadowed by the contrast of Hume and Kant. 251 SEBASTIAN GARDNER Naturalistic accounts ground art in human nature. The contingent ﬁttedness by nature of certain objects to our minds proposed by Hume provides the starting point. To exhibit the point of art, it needs to be shown how works of art engage fundamental and important mental activities (Dewey 1934). Psychoanalysis provides one example of this approach (Wollheim 1987). The question of the point of art is therefore answered by saying that art is natural: art exists because it is natural for us to create and appreciate it. As such, art is a ‘form of life’ (Wollheim 1980), as necessary a component of human exis- tence as language, culture and politics. The naturalistic view afﬁrms that art responds to the psychological needs of human beings, but it denies that the functions which art per- forms could be speciﬁed, or fulﬁlled, through other means. When art is pictured as a nec- essary part of the human order, and the appreciation of art as a natural component of human well-being, the point of art merges with that of human life itself. The naturalistic view of art therefore qualiﬁes as a ﬁfth theory of art, in addition to those considered earlier (it may, but need not, be formulated so as to incorporate the expression theory). By showing how the various dimensions of art map on to psycho- logical processes, it accounts for the unity of art in terms of the unity of the mind, and locates the essence of art with reference to human psychology. Metaphysical accounts of the point of art are, of necessity, more speculative and less pinned down to the empirical features of art than naturalistic accounts. They will also, of course, depend explicitly on a general philosophical outlook, which naturalistic accounts do not need to do. This does not, or should not, mean that they simply squeeze art into a preformed metaphysical system: they may, on the contrary, allow art a role in forming the system itself. This is what we ﬁnd, in different ways, in Kant, Hegel, Schopenhauer, Schiller and Nietzsche (German idealism has been notably more accommodating to art than any other philosophical tradition). The basic demand to which metaphysical accounts of art answer – and which nat- uralistic accounts cannot properly satisfy – derives from what might be called the ‘trans- ﬁgurative’ aspect of our experience of art: the sense that the transformation of reality which art effects, and which locates in it a kind of value that we ﬁnd consoling, is more than a fanciful embellishment. Tragedy exhibits the transﬁgurative power of art most clearly. The aspiration to achieve, through art, a justiﬁcation of the world ‘as an aes- thetic phenomenon’, as Nietzsche (1993: 32) put it, is a further part of the legacy of Romanticism, and has not disappeared from the demands that we put on art. Further Reading As an introduction to aesthetics, Richard Wollheim’s short Art and its Objects (1980), which offers both an overview and a distinctive approach to the subject, and Malcolm Budd’s Values of Art (1995) are both excellent. Other, plainer introductions are Charlton (1970) and Sheppard (1987). Hanﬂing (1992) contains a series of essays speciﬁcally written as a uniﬁed introduction to aesthetics, but of sufﬁcient length to present many arguments in detail. It is important to have a grasp at ﬁrst-hand of the classic writings in aesthetics, few of which make difﬁcult reading. Hofstadter and Kuhns (1964) is a ﬁrst-rate, currently available anthology. Carritt (1931) is more comprehensive but the extracts are shorter. Two other excellent antholo-	Blackwell
gies that also serve this end, and intersperse selections from the classics with modern writings, are Dickie and Sclafani (1977) and Rader (1979). Both are organized thematically and contain helpful bibliographies (as does Wollheim 1980). The historical development of aesthetics is traced 252 AESTHETICS by Beardsley (1966), which is also useful as a reference book. Cooper (1992) contains helpful entries on most ﬁgures and topics. A number of books may be singled out as offering more complex discussions with the stress on a particular topic or perspective. Danto (1981) is stimulating and addresses many problems, but with the accent ﬁrmly on contemporary art. Budd (1985) gives detailed and lucid analytical treatment of some central aesthetic theories. Wollheim (1974) brings issues in aesthetics in rela- tion to philosophy of mind and psychoanalysis. Scruton (1974) also emphasizes the contribu- tion of philosophy of mind, together with that of philosophy of language; Scruton (1983) relates aesthetics to cultural issues. Goodman (1976) has been highly inﬂuential. Savile (1982) focuses on questions of value, in the idealist tradition. Walton (1990) attempts to provide a general theory of artistic representation. Important articles are collected in Dickie and Sclafani (1977), Barrett (1965), Margolis (1978) and Osborne (1972). With respect to particular topics in aesthetics, as broken down in this chapter: Hume needs no commentary, but Kant’s Critique of Judgement is extremely difﬁcult, and it is best to read either Kemal (1992) or McCloskey (1987) alongside. Three comprehensive treatments of aesthetic experience and judgement are Beardsley (1982), Scruton (1974), and Mothersill (1984). Deﬁnitions of art are explored in Davies (1991). Theories of art, and accounts of its point, are best approached through the classic writings. For detailed treatment of the debates surrounding each of the dimensions of art, Schier (1986) deals with pictorial representation, Budd (1985) with expression, and Newton-de Molina (1984) with literary meaning and interpretation. References Aristotle 1987: Poetics (translated by S. Halliwell). London: Duckworth. Barrett, C. (ed.) 1965: Collected Papers on Aesthetics. Oxford: Blackwell. Beardsley, M. C. 1966: Aesthetics From Classical Greece to the Present. Tuscaloosa: University of Alabama Press. —— 1982: The Aesthetic Point of View. Ithaca, NY: Cornell University Press. Bell, C. 1914: Art. London: Chatto and Windus. Budd, M. 1985: Music and the Emotions. London: Routledge and Kegan Paul. —— 1995: Values of Art: Pictures, Poetry and Music. Harmondsworth: Penguin Books. Carritt, E. F. 1931: Philosophies of Beauty. Oxford: Clarendon Press. Charlton, W. 1970: Aesthetics. London: Hutchinson. Collingwood, R. G. 1937: The Principles of Art. Oxford: Oxford University Press. Cooper, D. (ed.) 1992: The Blackwell Companion to Aesthetics. Oxford: Blackwell. Croce, B. 1992 [1902]: The Aesthetic as the Science of Expression and of the Linguistic in General (translated by C. Lyas). Cambridge: Cambridge University Press. Danto, A. 1981: The Transﬁguration of the Commonplace. Cambridge, MA: Harvard University Press. Davies, S. 1991: Deﬁnitions of Art. Ithaca, NY: Cornell University Press. Dewey, J. 1934: Art as Experience. New York: Putnam. Dickie, G. 1984: The Art Circle. New York: Haven. Dickie, G. and Sclafani, R. J. (eds) 1977: Aesthetics: A Critical Anthology. New York: St Martin’s Press. Gombrich, E. H. 1960: Art and Illusion, 2nd edn. New York: Pantheon. Goodman, N. 1976: Languages of Art. Indianapolis: Hackett. Hanﬂing, O. (ed.) 1992: Philosophical Aesthetics. Oxford: Blackwell in association with the Open University. 253 SEBASTIAN GARDNER Hanslick, E. 1986 [1854]: On the Musically Beautiful (translated by G. Payzant). Indianapolis, IN: Hackett. Hegel, G. W. F. 1993 [1820–9]: Introductory Lectures on Aesthetics (translated by B. Bosanquet). Harmondsworth: Penguin Books. Hofstadter, A. and Kuhns, R. (eds) 1964: Philosophies of Art and Beauty. New York: Harper and Row.	Blackwell
Hopkins, A. 1982: Talking About Music. London: Pan. Hume, D. 1965 [1757]: Of the standard of taste. In ‘Of the Standard of Taste’ and Other Essays. New York: Bobbs-Merrill. Janaway, C. 1995: Images of Excellence: Plato’s Critique of the Arts. Oxford: Oxford University Press. Kant, I. 1987 [1790]: ‘Critique of aesthetic judgement’, Part I of Critique of Judgement (translated by W. S. Pluhar). Indianapolis: Hackett. Kemel, S. 1992: Kant’s Aesthetic Theory: An Introduction. London: Macmillan. Kivy, P. 1989: Sound Sentiment. Philadelphia, PA: Temple University Press. Langer, S. 1942: Philosophy in a New Key. Cambridge, MA: Harvard University Press. Leavis, F. R. 1986: Valuation in Criticism and Other Essays. Cambridge: Cambridge University Press. McCloskey, M. 1987: Kant’s Aesthetics. London: Macmillan. Margolis, J. (ed.) 1978: Philosophy Looks At the Arts. Philadelphia, PA: Temple University Press. Meager, R. 1970: Aesthetic Concepts. British Journal of Aesthetics, 10, 303–22. Moore, G. E. 1984 [1903]: Principia Ethica. Cambridge: Cambridge University Press. Mothersill, M. 1984: Beauty Restored. Oxford: Clarendon Press. Newton-de Molina, D. (ed.) 1984: On Literary Intention. Edinburgh: Edinburgh University Press. Nietzsche, F. 1993 [1871]: The Birth of Tragedy (translated by S. Whiteside). Harmondsworth: Penguin Books. Osborne, H. (ed.) 1972: Aesthetics. Oxford: Oxford University Press. Passmore, J. 1991: Serious Art. London: Duckworth. Peacocke, C. 1987: Depiction. Philosophical Review, 96, 383–409. Plato 1955: Republic (translated by H. D. P. Lee). Harmondsworth: Penguin Books. Rader, M. (ed.) 1979: A Modern Book of Esthetics, 5th edn. New York: Holt, Rinehart and Winston. Savile, A. 1982: The Test of Time. Oxford: Clarendon Press. Schaper, E. (ed.) 1983: Pleasure, Preference and Value. Cambridge: Cambridge University Press. Schier, F. 1986: Deeper Into Pictures. Cambridge: Cambridge University Press. Schiller, F. 1989 [1793–5]: On the Aesthetic Education of Man (translated by E. Wilkinson and L. A. Willoughby). Oxford: Clarendon Press. Schopenhauer, A. 1969 [1819]: The World as Will and Representation, 2 vols (translated by E. F. J. Payne). New York: Dover. Scruton, R. 1974: Art and Imagination. London: Methuen. —— 1979: The Aesthetics of Architecture. London: Methuen. —— 1983: The Aesthetic Understanding. London: Methuen. —— 1997: The Aesthetics of Music. Oxford: Clarendon Press. Sheppard, A. 1987: Aesthetics. Oxford: Oxford University Press. Sibley, F. 1959: Aesthetic Concepts. Philosophical Review, 68, 421–50. —— 1965: Aesthetic and Non-aesthetic. Philosophical Review, 74, 135–59. Sibley, F. and Tanner, M. 1968: Symposium on ‘Aesthetics and Objectivity’. Proceedings of the Aristotelian Society, supplementary volume 42, 31–72. Tilghman, B. R. 1984: But is it Art? Oxford: Blackwell. Tolstoy, L. 1930 [1898]: What is Art? (translated by A. Maude). Oxford: Oxford University Press. Walton, K. 1990: Mimesis as Make-Believe. Cambridge, MA: Harvard University Press. Wittgenstein, L. 1978: Lectures and Conversations on Aesthetics, Psychology and Religious Belief. Oxford: Blackwell. 254 Wollheim, R. 1974: On Art and the Mind. Cambridge, MA: Harvard University Press. —— 1980: Art and its Objects, 2nd edn. Cambridge: Cambridge University Press. —— 1987: Painting as an Art. London: Thames and Hudson. AESTHETICS Discussion Questions 1 What makes a judgement aesthetic? 2 Are the aesthetic qualities of an object any less objective than its colour? 3 Can arguments about aesthetic value be resolved, and what does it signify if they cannot? 4 Can it be shown, or must it merely be assumed, that aesthetic judgements differ from gustatory preferences? 5 In what ways, if any, can reasons be given for aesthetic judgements? 6 Are there any limits on what can be an object of aesthetic attention? 7 In what respects are aesthetic judgements like moral judgements, and in what respects are they unlike? 8 What can criticism of art hope to contribute to our appreciation of art?	Blackwell
9 In what respects can one’s relation to works of art be compared with one’s relations to other people? 10 Does aesthetic experience presuppose an aesthetic attitude? 11 Is it plausible to claim that beauty is the fundamental concept in aesthetics? 12 Does it matter if art cannot be deﬁned? 13 What is the relation between form and content in art, and does either have greater importance than the other? 14 To what extent is it appropriate and proﬁtable to compare art with language? 15 Does the concept of resemblance have any role to play in elucidating pictorial representation? 16 Can it be explained how music expresses emotion? 17 What assumptions are needed to uphold the claim that literary works have objective meanings, and are those assumptions acceptable? 18 Must each literary work have one correct interpretation? 19 Does the meaning of a literary work reside in the mind of its author, in the text itself, or in neither of these? 20 Is the correct interpretation of a literary work the one that makes it mean most to us? 21 Does the study of literature need to be guided by literary theory? 22 Does psychology have importance for aesthetics? 23 Do we need art? 24 Is it ever legitimate to view art as serving moral or political ends? 25 Is a paradox involved in the enjoyment of tragedy and, if so, how should it be resolved? 26 In what sense might it be true that art is a social rather than individual phenomenon? 27 Does emotional response to ﬁction require what Coleridge calls a ‘willing suspen- sion of disbelief ’? 28 Is it more than just a contingent truth that art has a history? 255 SEBASTIAN GARDNER 29 How can one be moved by the plight of Anna Karenina, when one knows her to be merely ﬁctional? 30 What distinguishes metaphor from nonsense? 31 Does survival of the test of time provide a criterion for artistic achievement? 32 Is photography a form of art? 33 Can it be maintained that any one form of art is superior to the others? 34 Is architecture ‘frozen music’? 35 Can anything be made of the thought that music has ‘metaphysical meaning’? 36 What makes a painting artistically signiﬁcant? 256 8 Political and Social Philosophy DAV I D A RC H A R D Social and political philosophy has experienced a dramatic revival in recent decades. This chapter considers the background of work in the years after 1945, before examining the writings on justice of the main architects of this revival, John Rawls and Robert Nozick. Subsequently discussed are the ideal of equality, the problem of pluralism and the prin- ciple of neutrality, the communitarian, feminist and Marxist critiques of liberalism, and the signiﬁcance of community. The ﬁnal section considers the nature of political philosophy and its relation to politics. 1 Introduction In a now celebrated phrase, Peter Laslett announced that ‘for the moment, anyway, political philosophy is dead’. He did so in his 1950 introduction to the ﬁrst collection of essays entitled Philosophy, Politics and Society. The sixth series was published in 1992, and the intervening years show clearly enough that the reports of death were greatly exaggerated, or at least that the moment was merely a passing one. The last ﬁfty years, and the last thirty in particular, have witnessed a quite astonishing regeneration of social and political philosophy, understood in Laslett’s own terms as the concern of philosophers with ‘political and social relationships at the widest possible level of generality’ (Laslett 1950: vii). The work has displayed a range of views, fertility imagination, rigour of argumentation, care in critical analysis, and concern to of address contemporary problems which are ample testimony to the considerable strengths of philosophy in the English-speaking world. In the 1950s these strengths were far more selectively deployed. The study of politics was, to simplify greatly, divided between the empirical social sciences and	Blackwell
philosophy. Exponents of the former were optimistic that political science could offer value-free but well-founded explanatory theories of its object. For its part philosophy disdained substantive political evaluation and restricted itself to the logical and linguistic analysis of political discourse. The title of Daniel Bell’s 1960 book, The End of Ideology, concisely deﬁned an his- torical moment, and its subtitle, ‘On the Exhaustion of Political Ideas in the Fifties’, spelled out what was intended. Bell’s text was a series of sociological essays about DAVID ARCHARD America, but its grander claims carried further aﬁeld. The 1930s and 1940s testiﬁed to the sheer awfulness of the practical realization of ideologies like fascism and Stalinism which made simplifying and absolute claims to truth about humanity, history and reason. These years also reconciled intellectuals to the virtues of a more moderate and modest political programme. Capitalism could be tamed by a welfarist state the economy mixed, and fundamental freedoms adequately protected by democratic constitutionalism. There was no further need for ideologies in the sense of millenarian visions of utopia yet to be realized. The exhaustion of ideologies in this narrow sense seemed also to betoken a disenchantment with politics in general, a retreat from the traditional concern of political philosophy to prescribe the good society. This was mirrored in the work of philosophers. The 1950s was the high tide of CONCEPTUAL ANALYSIS (pp. 2–3), and in an emblematic text, The Vocabulary of Politics, T. D. Weldon (1953) characterized the enduring questions of political philosophy as mis- placed, resting on false assumptions and the misuse of unanalysed fundamental terms. Correctly understood, these questions are merely ‘confused formulations of purely empirical difﬁculties’ (ibid.: 192). It is proper for philosophers to analyse the language and foundational claims of ideologies, but the evaluation of political states of affairs and institutions is more appropriately reserved for politicians and political scientists. However, it would be a mistake to over-simplify. From 1945 into the 1960s there was serious and important work in political philosophy. Karl Popper (1945) sought to rebut totalizing political theories such as those of PLATO (chapter 23) and MARX (chapter 34). But he did so by defending an ideal of the ‘open society’, and the politics of piecemeal reform rather than wholesale social engineering. F. A. von Hayek was entering upon his lifelong defence of a position which is echoed in important current views (Gray 1984). Hayek afﬁrmed the ideal of justice as individual freedom under the LAW (chapter 13), and defended the idea that the economic and social world displays a spontaneous, imper- sonal pattern. It is simply inappropriate, and ultimately destructive of freedom, for governments to impose upon this world according to a redundant ideal of ‘social’ justice. The work of Isaiah Berlin is especially noteworthy, for he anticipates some major themes in the philosophical liberalism that is now dominant. For Berlin, a plausible politics must start from a recognition of a plurality in valid yet conﬂicting ends of life. There is no single ideal of the good life, and the proper goal of politics is to make possible the pursuit by individuals of their several ends (Kocis 1980). Berlin (1958) famously distinguished two senses of liberty, one characterized as the absence of obstructions, and the other characterized in terms of self-mastery. He argued for the former, ‘negative’ liberty, as the ‘truer and more humane ideal’ than the latter, ‘positive’ liberty. At the same time Berlin urged a clear separation of the notions of liberty and equality. He conceded that the degree of social inequality may determine	Blackwell
the conditions that make liberty worth having, but insisted that these conditions do not enter into the deﬁnition of liberty. 2 John Rawls and Robert Nozick on Justice These various writings are important, but it is those of John Rawls that are now taken to signify the beginning of a distinctly new era. The publication of Rawls’s A Theory of 258 POLITICAL AND SOCIAL PHILOSOPHY Justice (1972) has proved a landmark, radically changing the character of English- speaking political philosophy and supplying the ineliminable background against which all subsequent discussion has been conducted. ‘Political philosophers now must either work within Rawls’s theory or explain why not’ is ﬁtting tribute from one whose own views are at a great distance from Rawls’s (Nozick 1974: 183). A Theory of Justice is much lauded but it is also now an extensively criticized text. Rawls’s presuppositions, methodology and conclusions have all been subjected to such extensive criticism that it is important to be clear how and why his text set the scene. First, A Theory of Justice is a work of substantive political philosophy. Although Rawls engages in the analysis of concepts, this serves rather than substitutes for the evaluation of political institutions. A Theory of Justice is more. It defends a single, uniﬁed vision of the good society – one that is recognizably liberal. Rawls gives philosophical voice to a doctrine of democratic constitutional liberalism which can with merit be urged as the only feasible theory for contemporary Western society. Second, Rawls seeks to combine a recommendation of the ideal polity with a plausible account of HUMAN NATURE (pp. 672–3). This involves not just a theory of human MOTIVATION (p. 392) taken from the SOCIAL SCIENCES (chapter 12). It also – and less noticed – comprises a view of the manner in which citizens can be brought to recognize and accept as both fair and realistic the formal terms of their coexistence. Third, Rawls has provided political philosophy with an agenda. He does not just offer a theory of justice. He supplies an account of the priority of individual liberty, the defensible limits of egalitarianism, and the rule of law. Fourth, Rawls’s work puts contemporary political philosophy back into contact with its own history. The work and concerns of past writers, such as HOBBES (chapter 28), LOCKE (chapter 29), Rousseau, HUME (chapter 31), MILL (chapter 35) and KANT (chapter 32), are given renewed pertinence. Fifth, A Theory of Justice is nevertheless a response to the particular circumstances of modernity. It is not just a principled afﬁrmation of the virtues of liberalism against the postwar background of disillusionment with more extreme ideologies. It can be seen as an attempt to afﬁrm these virtues in the context of a more local crisis, that is the challenge posed to the ideal of American democratic constitutionalism by the civil rights movement and the war in Vietnam. More generally again, Rawls, as did Berlin, insists that value pluralism is an unavoidable feature of modern existence to which politics must adequately respond. The ideas of A Theory of Justice can best be laid out by answering a number of broad questions. First, why is a theory of justice needed? Because publicly agreed terms of social cooperation are both necessary and possible. Individuals beneﬁt from living and working together so long as they can be assured that social existence is well-ordered and stable. Yet there is a predisposition to conﬂict inasmuch as individuals want differ- ent and not necessarily compatible things out of their life together. Agreed PRINCIPLES (pp. 733–6) are needed to regulate interaction, and determine the proper division of beneﬁts and costs among the members of society. Such principles are possible to the extent that individuals can see them to be necessary, and their particular terms can be	Blackwell
agreed under appropriate conditions. 259 DAVID ARCHARD The Primacy of Justice Justice has primacy in three senses. It is the ﬁrst virtue of social institutions. This does not mean that society cannot display other virtues, only that without at least the assur- ance of justice these would have little or no value. Second, truth is a demand made of a system of thought in so far as it relates to reality. Justice is demanded of a polity in so far as it must relate to the real world. The circumstances of justice are those ﬁrst noted by David Hume, namely limited benevolence and moderate scarcity. Humans are not well- enough disposed to make sacriﬁces for others, and material resources are not available in sufﬁcient quantity to make formal agreement on terms of cooperation unnecessary. Rawls adds to this understanding of circumstances the facts of modern value pluralism. Individuals have different but not necessarily compatible aims. My pursuit of my life’s ends will conﬂict with your pursuit of yours. Agreement on how to reconcile such a conﬂict is necessary but possible only if no one set of individual values is afﬁrmed over the others. Third, any theory of justice speciﬁes and guarantees an equality of citizenship which supplies an assurance that no individual may be sacriﬁced for society’s greater good. Justice has primacy in that this assurance is not to be compromised. What is a theory of justice? It is a theory of the publicly agreed, ﬁnal principles which deﬁne the fundamental terms of social co-operation. It is a theory of those principles which regulate the institutions, the ‘basic structure’ of society. It does not prescribe par- ticular outcomes; nor does it provide a criterion for evaluating actions or the charac- individuals. It is a theory of social justice, which must also ter and dispositions of coincide with our considered and reﬂective MORAL JUDGEMENTS (pp. 225–7). Rawls’s notion of ‘reﬂective equilibrium’ allows for, indeed encourages the possibility of, mutual adjustment in both theory and judgements. Yet any theory of justice must capture what is essential to and shared by different accounts of justice, namely that ‘institutions are just when no arbitrary distinctions are made between persons in the assigning of basic rights and duties and when the rules determine a proper balance between competing claims to the advantages of social life’ (Rawls 1972: 5). What then is A Theory of Justice? Rawls’s theory assures citizens equal LIBERTY (pp. 762–3), and a distribution of all other goods that maximizes the expectations of the least well off. More formally and completely: First Principle Each person is to have an equal right to the most extensive total system of equal basic liberties compatible with a similar system of liberty for all. Second Principle Social and economic inequalities are to be arranged so that they are both: (a) to the greatest beneﬁt of the least advantaged, consistent with the just savings principle, and (b) attached to ofﬁces and positions open to all under conditions of opportunity. (Rawls 1972: 302) fair equality of The ﬁrst part of the second principle is familiarly known as the ‘difference principle’, and the two principles are in lexical order. That is to say that inequalities regulated 260 POLITICAL AND SOCIAL PHILOSOPHY by the second principle are only permitted when equality of liberty under the ﬁrst principle has been guaranteed. Provided that these principles apply or very nearly apply to the basic institutions of society then that society is just. And individuals have a duty to abide by the rules of a just society. Although Rawls offers a CONTRACTUALIST (pp. 672–7) argument for the rules themselves, the duty to abide by them is not itself contractually based. In part three of A Theory of Justice Rawls offers an account of the acquisition of a sense of justice. This grows out of the basic attachments and relationships which constitute any	Blackwell
society: family, friendship and broader associations. There is in the well-ordered society a concordance of fair rules and a sense of justice. Citizens can recognize that the rules are just and acknowledge their duty to abide by them. Fairness and feasibility coincide. Lastly, A Theory of Justice is a theory of THE RIGHT AND THE GOOD (pp. 598–601). The just society is not one which realizes the good life of its community. It is one which permits its members to pursue their own conception of the good under certain condi- tions. First, the pursuit of the good is constrained by the right, that is the publicly agreed principles of justice. Second, these principles determine the distribution of primary social goods – rights, liberties, powers and opportunities, income and wealth – which are the desired prerequisites of any particular pursuit of the good. Third, people pursue different ideals of the good life. ‘Human good is heterogeneous because the aims of the self are heterogeneous’ (Rawls 1972: 554). It would be a disastrous mistake for a polity to try to impose upon its members any one particular ideal of life. Rawls has continued to believe that, in any modern society, there will be a plurality of conscientiously sought ends, and that any state, presuming or prescribing to the contrary, would have to employ extensive coercion to secure its single vision. The ﬁnal question to be answered is, Why A Theory of Justice? This asks what justiﬁ- cation Rawls offers for his own principles. Rawls famously employs a contractualist methodology: the principles of justice are those that rational, self-interested, but mutu- ally disinterested individuals would choose in an original situation speciﬁed chieﬂy in terms of the parties’ selective knowledge. The contractualist method has been the subject of a great deal of criticism: why should a hypothetical contract bind? Why would individuals choose as Rawls claimed they would? Would not individuals choose differently in a different original position? Moreover Rawls himself has subsequently denied that the contractarian argument has independent justiﬁcatory force and views it only as a heuristic device which serves to illustrate the force of a claim whose real justiﬁcation lies elsewhere. However, the question then presses of what does justify the principles of justice, and it presses the more since Rawls has become concerned with the issue of how a just society can be well-ordered, that is be viewed as legitimate by its citizens. Before considering this question it is necessary to examine a fundamental, and inﬂuential, critique of Rawls’s understanding of justice. This is to be found in Robert Nozick’s Anarchy, State and Utopia (1974), which is rightly paired with Rawls’s A Theory of Justice when the contemporary resurgence of political philosophy is discussed. The foundations of Nozick’s arguments are RIGHTS (pp. 690–1) and OWNERSHIP (pp. 690–1). The rights possessed by individuals are funda- mental, and deﬁne a moral space surrounding each person whose invasion constitutes a wrong. Following Locke, Nozick holds the basic rights to be those to life, liberty, and 261 DAVID ARCHARD ‘estate’ or property. For Nozick, these rights are ‘side-constraints’ upon action, and have inﬁnite moral weight. That is, no amount of good consequences, including even an overall diminution of rights violations, can justify a single violation of one of these rights. Nozick does not offer a systematic defence of his understanding of rights. At most he claims the underlying justiﬁcation to be the Kantian view of INDIVIDUALS AS ENDS (p. 736), leading separate, different, consciously and deliberatively shaped lives, which should not be sacriﬁced or used as means to others’ ends. Nozick is egalitarian to the extent of holding that all humans equally possess these fundamental rights. He also endorses an initial equal distribution of ownership in	Blackwell
respect of our selves. Each owns his or her own body, and its powers, capacities and abilities. Nozick believes that through an exercise of this initial self-ownership further entitlements to bits of the world are generated. He considers only one form of argu- ment to show how this generation of entitlements might occur. This is John Locke’s famous claim that one acquires rights to that with which one has mixed one’s labour. Unfortunately Nozick devotes four incisive pages to expose seemingly unanswerable difﬁculties in the Lockean account (Nozick 1974: 174–8). This seems to leave Nozick with his own problem of unfounded entitlements, and so some critics have charged. But Nozick is appealing to the plausible background assumption that unowned objects are there to be owned, and that some sort of fruitful exercise of one’s own powers grounds an entitlement to ownership of these objects. That is, provided certain conditions upon legitimate acquisition are met. Entitlements What occupies Nozick is less the process of acquisition than the limits which may be set upon its scope. Nozick borrows again from Locke and urges the acceptance of one fun- damental proviso: ‘A process normally giving rise to a permanent bequeathable property right in a previously unowned thing will not do so if the position of others no longer at liberty to use the thing is thereby worsened’ (ibid.: 178). Nozick’s concern is to show that this proviso can be met by an established distribution of property entitlements, when one considers that there may no longer be any unowned objects for individuals to acquire. His answer is that individuals unable to appropriate are nevertheless better off living under a system permitting private property than they would be if no original appropria- tions had been permitted. The account of entitlements is completed by acknowledging that those who hold bits of the world are entitled freely to transfer them if and to whomsoever they choose. It follows that those who receive such freely transferred holdings are themselves now entitled to hold them. Thus everything which is not unowned is legitimately owned if acquired justly (by some process which does not violate the Lockean proviso) or freely transferred from one who justly acquired it. A just set of holdings is just that set of holdings which came about in the right way. In part one of Anarchy, State and Utopia Nozick justiﬁes a minimal state. He does so by constructing a plausible tale of how a state of nature might have given way to a state, that is an organization legitimately claiming a monopoly on the use of force within a given territory, without at any stage violating rights. The tale turns on the evolving 262 POLITICAL AND SOCIAL PHILOSOPHY competition between private agencies formed to protect their clients, with one eventu- ally assuming a monopolistic role and adequately compensating the others for its usurpation of their functions. Part two of Anarchy, State and Utopia is devoted to a defence of the view that such a state, restricted to preventing force, theft and fraud, and enforcing contracts, is morally sufﬁcient. Any greater role for the state is illegitimate. Of course an obvious reason for a greater role would be that people’s holdings need to be redistributed in accord with a theory of justice. Nozick’s own theory of justice does not require any such action by government. His task is then to show that redistributive theories of justice are mistaken, and what they require of the state illicit. The distinction Nozick draws is between his own entitlement theory of justice, and patterned or end-state theories. His is a historical theory since the justice of a set of holdings is given by the propriety of the history which led up to it. End-state theories are unhistorical, specifying that a distribution must conform to a prescribed structure.	Blackwell
More particularly, patterned theories prescribe a distribution of holdings according to some natural attribute or ordered set of such attributes, such as intelligence or effort. Nozick’s critique of non-historical theories is both general and speciﬁc. At a general level he insists that no patterned distribution can be maintained without persistent and serious violations of the right to liberty. His reasoning is simple. Voluntary transfers of those holdings initially assured under any patterned distribution will subvert the pattern and transgress the principle underlying the pattern. Such transfers are easy to imagine, and appear eminently unobjectionable inasmuch as they can be simple, fans willingly consensual, bilateral exchanges. Nozick’s own famous example is of paying extra to see an especially talented basketball player, Wilt Chamberlain. Such transfers can only be prevented by denying individuals the right to do with their holdings as they choose. Surely no consistent theory of justice can both distribute holdings and deny the individuals to whom they come any effective control over them? Nozick’s speciﬁc criticisms are of Rawlsian and egalitarian theories of justice. The fundamental weakness of Rawls’s theory is that it proceeds as if individuals did not already have claims to ownership of themselves and bits of the world. A cake sponta- neously given to a group might be divided among its members in a way close to Rawls’s principles; a cake which I have baked or whose ingredients you have provided will be divided in quite another way. Rawls is simply wrong to discount the various entitle- ments individuals bring to any determination of who gets what. People, not least, are entitled to their natural assets, and, consequently, to any rewards that may ﬂow from these. Rawls is mistaken to think that these assets are a collective resource, or that the difference principle which reﬂects this assumption would be acceptable to the well-off who are required to make a comparatively greater sacriﬁce so that the worst-off beneﬁt. Nozick ﬁnds no justiﬁcation either for any egalitarian principle. A demand for equality of material condition reduces to an unsubstantiated claim that the function of society is to meet the needs of its members, and simply discounts the fact that things are already attached to particular individuals. Equality is not required for self- esteem, which in fact thrives on comparable differences. At base the demand for equality is driven by envy. Nozick’s theory of justice permits the government no interventionist role beyond the rectiﬁcation of injustices by the terms of the entitlement principle. It is ironic then that, 263 DAVID ARCHARD given the logistic difﬁculties in tracking back and forward between past wrongs and present holdings, Nozick should suggest the difference principle as a rough rule of thumb for rectifying injustices (ibid.: 231). Otherwise holdings must be left as they have come legitimately to stand. That this may result in great disparities of wealth and life prospects is regrettable but no injustice. The rich may choose to be philanthropists but they do no wrong in not being. And it would violate their rights to require that they assist the less fortunate. This seems harsh but it is not obviously wrong-headed. Liberal critics chide Nozick for offering a ‘libertarianism without foundations’, but this may not be quite true. Nozick’s foundations are attractive ones. Individuals do appear to have rights in the sense that it would be fundamentally wrong to do certain things to them in the name of a greater social good. Individuals do also have a claim to ownership of their selves. That this is so can be shown by considering one’s reaction to the idea that bodily parts should be redistributed equitably or to beneﬁt the least fortunate in this respect. Do the blind have a claim upon at least one of the eyes of the sighted?	Blackwell
Criticism of Nozick is more appropriately directed at his rendering of these founda- tions and how he builds upon them. There is little warrant for construing rights in an absolute and exclusively negative manner, not least when appeal is made to Kantian underpinnings. Within a moral community, individuals in need may, on Kant’s own principles, have the right to contributions from those who can help them. It is an impos- sibly stringent requirement of rights merely that they rule out invasions of our moral space and do nothing to sustain us in our satisfaction of basic needs. Nozick may be right to think that we, severally, own ourselves and may, in conse- quence and by some form of activity, come to make legitimate claims upon what is unowned. But, arguably, the conditions he sets for legitimate acquisition are too lax and too easily met in ways that favour free-market capitalism. It may very well be possible to concede an initial equality in self-ownership, and the legitimacy of some process of acquisition of unowned objects, but then to specify sufﬁciently tough conditions of legitimacy to ensure that a ﬁnal equality of holdings is ensured. Again, the ‘Lockean proviso’ need not be as easily satisﬁed by his favoured set of economic arrangements as Nozick assumes, especially once ambiguities in both the sense of ‘better off ’ and the baseline against which comparisons are made are clariﬁed. Nozick seems to presume throughout that ownership is full and exclusive ownership by individuals, and that anything less amounts to a violation of liberty rights. Yet Nozick forgets that, for Locke, individuals made claim through their labouring not upon what was unowned, but what was communally owned. In his Wilt Chamberlain example Nozick assumes that individuals within the patterned distribution have hold- ings with which they are free to do as they choose. All forms of redistribution, includ- ing taxation, are implausibly stigmatized as coercive interferences with freedom. Yet it is possible that individuals might choose to limit in advance departures from a favoured pattern of distribution, and do so in the name of a liberty-preserving equality of condition. In sum, Nozick’s theory is not so much libertarianism without foundations, as libertarianism with unwarranted conclusions. His case against anarchy may be accepted. His claim that state minimalism is utopian in a positive sense remains unproven. 264 POLITICAL AND SOCIAL PHILOSOPHY 3 Equality Both Rawls and Nozick are egalitarians to the extent that they are, in their different ways, committed to the view that human beings are entitled to an equality of regard or treatment in some fundamental respect. Rawls’s theory assures citizens equal liberty and Nozick holds that all humans equally possess fundamental rights. All contem- porary social and political philosophy afﬁrms that humans are entitled to equality of something, which suggests that the crucial question is not ‘Why equality?’ but ‘Equality of what?’ (Sen 1992). Of course a demand for equality in respect of some good is consistent with, indeed may demand, inequality in respect of some other good. This is true of Nozick’s libertarianism which claims that an equal distribution of individual liberty must lead to an unequal distribution of income and property. A demand for equality in some good may take priority over other values, as is the case with Rawls’s insistence upon the lexical priority of equal liberties. But although Rawls’s ﬁrst principle of justice formally guarantees equal citizenship, the difference principle tolerates social and economic inequality. Rawls sees no incon- sistency here, though he concedes – as did Berlin – that differential access to resources will qualify the worth of equal liberty. The more you have the more you can make of a freedom shared equally with others in society. Socialist critics of liberalism have always	Blackwell
insisted that this is unsatisfactory, arguing that real equality of freedom demands an equalization of resources. Some feminists have also argued that the public or legal equality of the genders is undermined by, and yet serves to disguise, a fundamental inequality of social power. This characterizes all male–female relationships within patriarchal society, and its elimination will require more far-reaching reform of society than simply instituting a principle of equal citizenship. Walzer’s (1983) argument is important in this context. He suggests that there are different spheres of justice, each being speciﬁed by a set of goods to be distributed and a consequent principle of their fair distribution. The unequal distribution of some good need not in itself be unjust, but if this distribution determines the distribution of some other good then that is unfair. It is not necessarily wrong that some people have more money than others; it is if their greater wealth buys them political power, ofﬁce or greater personal health. Walzer’s approach highlights the fact that our favoured account of equality depends on what it is that we wish to see equally distributed. Contemporary egalitarianism defends three broad ﬁelds of application of equality. For a welfare egalitarian the ideal is a ‘condition of equal well-being for all persons at the highest possible level of well-being’ (Landesman 1983). The central problem for welfare egalitarianism is that it seems com- mitted to taking account of those pleasures some humans may take from seeing others do less well, and is also committed to the satisfaction of some acquired expensive tastes. Resource egalitarianism assigns to each individual a bundle of goods which is envied by no other individual. The central problem for this version of egalitarianism is that natural assets (skills, talents and abilities) are unequally distributed and in consequence people beneﬁt differentially from their use of equally distributed resources. Yet to include personal talents among resources which are equalized means that the talented are, in effect, the slaves of the untalented. 265 DAVID ARCHARD Contemporary egalitarians are now inclined to agree that individuals should not be compensated for the effects of free choices (such as, obviously, choosing to develop an expensive taste), but should be compensated for those factors affecting them which are due to ‘brute luck’ (such as, obviously, a handicap). On Dworkin’s (1981) inﬂuential argument, a principle of liberal equality should be ‘endowment-insensitive’ but ‘ambition-sensitive’. That is, it should permit individuals’ lives to ﬂourish or founder as a result of the choices they make, but not in consequence of their natural or social endowments. Such a principle may not be easy to render determinate and relies upon a speciﬁcation of the scope of free choice. Moreover it should be noted that individuals may not choose the circumstances which cause freely chosen preferences to be more or less expensive. Capability or opportunity egalitarians demand the equalization of the capacity to lead the life that the individual values or chooses to live. They thereby avoid the problem faced by resource egalitarians of how the same set of resources may, according to the circumstances, be differentially convertible into achievable standards of living. They also avoid the problems of welfare egalitarianism which neglects those aspects of good life which do not reduce to well-being and which cannot acknowledge the fact that some people, subject to enduring conditions of signiﬁcant inequality, may adapt their preferences (and states of well-being) to these conditions and not, in consequence, expe- rience a signiﬁcantly lesser degree of welfare. Capability egalitarians, however, must provide a ranking of capacities which cannot be purely quantitative – a life does not necessarily go better the more things one can do. But if such a ranking does not reduce	Blackwell
to one of opportunities for greater or lesser welfare, it is in danger of being an objective list resting upon a contentious account of the human good. Finally, any egalitarianism must answer criticisms of both practice and principle. Egalitarianism may simply be inconsistent with certain immutable features of human motivation. It may not be possible to combine, within the individual, the attitude of universal impartiality and personal partiality (Nagel 1991). Again, a society in which equality is guaranteed may lack the incentives necessary for maximizing the total social product and thereby improving the well-being of all. Egalitarianism must also meet Nozick’s challenge that the case for equality is unproven, and anyway trumped by more fundamental ideals, such as freedom or self-ownership: a prior commitment to liberty undermines any guarantee of equality, and, correlatively, equality can only be maintained by the denial of individual freedom. The disputed relationship between equality and liberty may prove to be the most enduring and pressing issue of political philosophy. The fact that we have thereby come full circle to the work of Berlin and Hayek certainly suggests an underlying continuity of concern in the subject. 4 Pluralism and Neutrality Rawls responded to criticisms of A Theory of Justice in a series of articles which culmi- nated in his second major book, Political Liberalism (Rawls 1993). Its concern is with the good order of a modern liberal democratic society. It does not defend a particular theory of distributive justice, although Rawls continues to think that his own two prin- 266 POLITICAL AND SOCIAL PHILOSOPHY ciples are suitable to regulate the basic structure of a well-ordered society. Whereas A Theory of Justice started from Humean circumstances of justice, Political Liberalism starts from the modern condition of value pluralism, namely that people entertain, for good reasons, different and probably incompatible comprehensive philosophical doctrines. Political Liberalism also repudiates the main arguments for the two principles of A Theory of Justice. The contractarian argument was only a ‘device of representation’; no meta- physical understanding of the self was or needs to be presupposed; and it is a mistake to think agreement to the principles of justice could rely on deep-lying acceptance of a broader moral doctrine of fairness. Nevertheless, Political Liberalism argues that a political conception of justice could command the support of an ‘overlapping consensus’ of various comprehensive doc- trines in the society whose basic structure it regulates. A just society of free and equal citizens could thus endure over time even though deeply divided in its basic religious, moral and philosophical outlooks. Some critics fear that Rawls’s theory only represents a pragmatic accommodation to the possibilities of consensus within particular societies. Yet Political Liberalism does not simply defend the politics of compromise and modus vivendi. It is an attempt to specify the terms of co-operation that can withstand the test of public and rational negotia- tion, commanding the freely given assent of equals. The worry should rather be that Rawls can no longer show why the well-ordered society must be one regulated by certain principles of justice, and vice versa. Rawls’s idea of an ‘overlapping consensus’ may be so indeterminate as to yield very many different outcomes. Or it may deliver a speciﬁc outcome by prescriptively stipulating what shall and shall not count as reasonable doctrines. Moreover, it may miss the real nature of negotiating political difference in its representation of views and disagreement as ‘reasonable’. The last part of A Theory of Justice sketched an account of the development in the citizens of the just society of the requisite sense of justice. It offered a theory of moral education through the relationships of association within that society. In Political	Blackwell
Liberalism Rawls forswears this kind of account. But to that extent any theory of political justice is ‘thinner’ and less plausible. Indeed there is a general problem of the basic liberal approach. Rawls’s conviction remains that, short of an unacceptable coerced unanimity, a domain of public agreement can be secured and clearly separated from the sphere of the heterogeneous private good. In this he remains faithful to the liberal vision of a society in which all equally and freely pursue their different lives constrained by public terms of fair coexistence. Yet such a vision may be impaled on the horns of a dilemma. Either the terms of public agreement are so insubstantial as to yield only an empty form of political community. Or they are made substantial only by violating the requirement that disputed understandings of the good be restricted to the private sphere. A liberal political order may demand a culture of and education in liberal values. The good of society may not be so readily separable from the good of its citizens’ private lives. Finally it may be held unreasonable to ask people to exclude their deeply held com- prehensive views from the terms of their political activity. Citizens are required to frame and articulate their demands in the language of a public reason which is political and independent of any particular comprehensive view. In a frequently invoked metaphor 267 DAVID ARCHARD individuals must come ‘naked to the public square’ by setting aside views that are deeply important to and perhaps even constitutive of their selves. Moreover a liberal political culture is cut off from the substantive comprehensive views that may historically have nourished it, leaving it deracinated and its citizens alienated. Rawls’s starting point in Political Liberalism, as it is now for many other political philosophers, is value pluralism. This should not be confused with relativism or scepti- cism about the good. Such pluralism is rather taken to be an inevitable result of the conscientious exercise by distinct individuals of a similar capacity for reasoning. An expectation of reasonable disagreement is an acknowledgement of an evident fact about modern life and a rejoinder to a long Enlightenment tradition which holds reason to generate convergence on the truth. Pluralism must not merely be expected but also tolerated because value monism can only be secured through coercive state interference with individual liberty of con- science, and because pluralism need not spell disaster if it can still provide the basis of harmonious social and political co-operation. The problem may be that the toleration of value pluralism presupposes some foundational value – such as the autonomy whose exercise by different individuals generates it. Yet if pluralism goes, as it were, all the way down there is no reason to think that any value can be privileged and not be the subject of reasonable disagreement between individuals. Value pluralism is honoured, arguably, if and only if the state, in its laws and policies, remains neutral between its citizens’ different conceptions of the good. This doctrine of ofﬁcial neutrality on the question of the good is viewed by many as deﬁnitive or constitutive of contemporary philosophical liberalism (Dworkin 1978; Ackerman 1980: 10–12). A distinction is normally drawn between neutrality with respect to the justiﬁcation or aim offered by government for its activity and with respect to its conse- quences or outcome, most preferring to understand neutrality in the ﬁrst sense. Thus, going back to a much earlier liberal thinker, John Locke defended political intolerance of Catholics (non-neutral in its consequences) not because Catholics were doctrinally in error (non-neutral in its justiﬁcation) but because Catholics, in owing allegiance to the foreign jurisdiction of Rome, represented a danger to the good order of the state (neutral in its justiﬁcation) (Locke 1689).	Blackwell
The manner in which political neutrality can be achieved is represented either negatively, as the bracketing out of whatever is the subject of disagreement, or positively, as the operation of a shared public political reason. Neutrality will be defended by appeal to the value of equality or that of individual autonomy. On the ﬁrst a state does not treat its citizens as equals if it favours one citizen’s views of the good over others. On the second it is more important that individuals lead the lives they see as good rather than be led to live the life that the state thinks of as good. In addition a doctrine of neutrality will be seen as a necessary protection against the excessive, and dangerous, exercise of state power over individuals. Value pluralism, and the associated principle of neutrality, stand opposed to perfec- tionism, the doctrine that there are speciﬁable human excellences such that some forms of human life are superior in themselves to others. Rawls rejects perfectionism, but some liberals have insisted that, nevertheless, his own account of human good is too thin, and that we can make judgements about the relative worth of different forms of existence. This need not be inconsistent with a commitment to egalitarianism and 268 POLITICAL AND SOCIAL PHILOSOPHY tolerance (Haskar 1979; Galston 1980). Moreover, there is a suspicion that Rawlsian liberalism is, at base, inconsistent. If it is strictly neutral then it cannot subscribe to any normative understanding of individuals, that they should, for instance, strive to be purposive, autonomous creatures. But such a view seems presupposed by Rawls. If, on the other hand, liberalism does include some foundational moral judgements about human beings, then it cannot reasonably claim objectivity for these judgements and refuse it for judgements about the good life. It also seems clear that a liberal democratic culture will ﬂourish to the extent that its citizens acquire and practise certain virtues, those for instance of tolerance, civility, respect for others and a willingness to make sacriﬁces for the common good. Now whether such virtues are learnt in childhood, or got through good habits, a liberal society must surely take the decision to encourage those social forms which facilitate the acquisition of such virtues. Autonomy and Liberalism In this context Joseph Raz (1986) is a revisionist liberal. He doubts whether liberty should be the central value, and defends instead the primacy of autonomy. His understanding of autonomy is that only a life chosen from among several moral options is autonomous. Further, Raz believes that, while not all forms of life are valuable, there may be several incompatible ones that are. Autonomy requires a plurality of moral life choices, and this in turn requires the creation and maintenance of social forms conducive to autonomy. The government has a duty to preserve these forms but, crucially, may do so in non-coercive ways. Subsidies and taxation can effectively render certain choices attractive or unattractive relative to others. Thus, Raz bites the bullet, holding that a liberal political culture should sustain its own core values, and not aspire to an implausible ideal of neutrality. Liberalism can be tolerant of diversity, interventionist and anti-perfectionist. In similar terms William Galston’s liberalism embraces and supports a set of distinctive liberal purposes that guide liberal public policy and shape liberal justice. These require the practise of and a civic education in liberal virtues, and the maintenance of a liberal public culture. Liberalism has a thick enough theory of the good life to be able to rule out certain practices and encourage others: ‘it is not the absence of an account of the good that distinguishes liberalism from other forms of political theory and practice. It	Blackwell
is rather a special set of reasons for restricting the movement from the good to public coercion’ (Galston 1980: 180). Liberalism is not neutral; rather its governance is not morally costly in its use of coercive state power. 5 Critics of Liberalism: Communitarianism, Feminism, and Analytical Marxism There are three broad movements deserving consideration for their critiques of liberalism: communitarianism, feminism and Marxism. 269 DAVID ARCHARD 5.1 Communitarianism The writers gathered under this title – Roberto Unger (1975), Michael Sandel (1982), Michael Walzer (1983), Charles Taylor (1985), Alasdair MacIntyre (1981, 1988) and Richard Rorty (1989, 1991) – are disparate and do not consciously subscribe to a common manifesto. It is more accurate to speak of family resemblances than a single, shared programme. In the respects in which they are all communitarian they offer not so much an alternative political view to liberalism as a criticism of its presuppositions. Communitarians invoke community in criticisms which are both normative and descriptive, although the distinct kinds of criticism are not always carefully distin- guished (Caney 1992; Taylor 1989). The central descriptive criticism concerns the nature of the self or individual, and charges liberalism with subscribing to an inadequately ‘thin’ understanding of the ‘self ’. The facts specifying the social and historical situation of each person constrain the kinds of self-understanding she can reach, and the choices she will make. The Rawlsian individual is not ‘embedded’ in any place or time; she is so emptied of substan- tial, individuating features as to make it difﬁcult to describe her as choosing a life. How can a self-less person be said to have any conception of the good or make choices of ends? This criticism may rest on a misunderstanding. Rawls’s concern is not to deﬁne ide- alized choosers of the good so much as to specify the relevant considerations entering into a jointly agreed determination of the public rules of fair co-operation. As he insists in his later work, his theory of justice is not metaphysical but political and assumes no particular understanding of the self. At the same time the criticism may be overstated. To claim that individuals are wholly deﬁned, and their identity completely constituted by their membership of some community at some historical moment, is effectively to deny them any kind of meaningful choice over their lives. It also seriously undermines any claim they might make to be MORAL AGENTS (chapter 6). And to the extent that the communitarian claim is qualiﬁed by phrases such as ‘to a large extent’ it is hard to see what distinguishes the communitarian from the liberal. The normative criticisms of communitarianism are threefold. First, the priority which Rawls accords the virtue of justice is alleged to derive from an impoverished understanding of political association. Rawls claims that justice is the ﬁrst virtue of social institutions, that it is required to deal with limited benevolence, moderate scarcity and modern value pluralism, and that it is needed to protect individuals from being sacriﬁced for the society’s greater good. According to communitarian critics, however, justice is only the ideal of societies which do not display community. Communities proper do not need to be just, and would not be communities if they felt such a need. Michael Sandel (1982) characterizes justice as a remedial virtue, binding up in the best way possible a second-best form of social co-operation. The sense in which justice has primacy for Rawls has already been indicated, and Sandel’s criticism appears misplaced. Resolution in the face of death is not less of a human virtue for being redundant in immortal creatures. Sandel does not claim, as have some Marxists for instance, that the circumstances of justice will disappear in the future. He says only that some communities do not display them. But these – most	Blackwell
notably the family – cannot be the models for society as a whole. The family has those characteristics which make its members benevolently disposed to one another – 270 POLITICAL AND SOCIAL PHILOSOPHY closeness, and affectivity rooted in natural relationships – because it is not simply a smaller kind of society but something quite different altogether. Societies cannot be familial communities. The second normative claim of communitarianism is that membership of a political community is a good which liberalism neglects, ignores, or whose sense it cannot successfully capture by its own terms. Political association is viewed by liberals and libertarians as an INSTRUMENTAL GOOD (pp. 216–18). It realizes the compromises necessary for individuals to derive mutual advantage from co-operating. There is no other sense of being together as citizens than is required to bring about this end. Rawls and Nozick both talk of community, or even communities, growing up within this framework. But these seem inessential and somehow added onto the basic terms of political association. This would not be so serious a criticism were it not for the further claim that a liberal theory of justice needs more. Sandel charges that Rawls defends a difference principle without foundations. Acceptance of this principle requires a willingness to see one’s natural assets as communally owned, and yet Rawls’s theory allows for no community which could lay such a claim to ownership. On Rawls’s account I see my talents as for others to derive beneﬁts from, yet there is no reason why I should see myself as joined to these others. Here we meet again the problem for liberalism of showing how a just society can also be well-ordered and whether political legitimacy must rest upon a sense of community which liberalism is incapable of supplying. The third normative communitarian claim is that what is good and just for individ- uals is deﬁned by the community to which they belong. Alasdair MacIntyre (1981, 1988) is associated with the view that an individual’s good is inseparable from the ROLE (p. 388), ofﬁce or social position he ﬁlls. Michael Walzer (1983) makes the further claim that, since the goods to be distributed have particular SOCIAL MEANINGS (pp. 384–90), the justice of their distribution is relative to these meanings. As these meanings have particular social and historical location, so do the associated principles of justice. However, the besetting danger of any appeal to the existence of distinct under- standings of the good and the just is RELATIVISM (pp. 395–7). This society is just by its own lights; that society is just by its own lights. And never shall the two be compared. Further, shared understandings determine not only what is just but what is unjust. Yet these particular judgements may violate what, plausibly, ought to be UNIVERSAL MORAL STANDARDS (pp. 733–6). Imagine a society by whose understanding of the master–slave relationship 20 lashes a day are fair and sufﬁcient. If a master adminis- ters 25 he is unjust and if only 15 he behaves generously. But surely any beating is wrong, and any ‘rules’ of slavery are unjust, whatever the members of the society believe. With regard to the communitarians’ normative claims the liberal will respond that liberalism does not dispute the value of community nor need it neglect its signiﬁcance. The liberal will press the communitarian to clarify precisely what sort of political pro- posals are distinctive of communitarians, worrying that, where they are spelled out, such proposals suggest an illiberal concession to whatever happens to be the shared understandings and practices of a particular time and place (Gutman 1985). At best communitarianism prompts liberalism to demonstrate how its constitutive values are 271 DAVID ARCHARD	Blackwell
consistent with the maintenance and reproduction of a good order which, arguably, needs a shared sense of community. 5.2 Feminism FEMINISM (chapter 20) offers two quite distinct kinds of criticism of philosophical liberalism. The ﬁrst concentrates on the silence of liberal political theory and, being consciously ad hominem, of its male theorists about women’s place in a just society. More particularly this silence is compounded by assumptions about what this place actually is and should remain. A formal commitment to the equality of all is gainsaid by an endorsement of patriarchalism, which is not always only implicit. Patriarchalism here means the doctrine of the subordination of women to male power. The Public and the Private The crucial assumption underpinning liberal patriarchalism is that to the distinction between public and private spheres of activity corresponds a difference of nature between men and women, and the roles for which these natures best suit them. The woman is conﬁned to the family where her biology equips her to bear, rear and care. The woman is thus doubly oppressed: excluded from the public, political sphere where the liberal principle of equality operates, and the inferior of her male provider within the private household. In terms of oppression related to the distinction between public and private spheres, Susan Moller Okin (1989) accuses Rawls of perpetuating liberal patriarchalism. He never broaches the issue of sexual justice, yet assumes both the continued existence of the institution of the family, and, more pertinently, the traditional sexual division of labour within this institution. Okin can reasonably claim that a family whose structure is deeply unjust cannot, as Rawls expects, be the appropriate institution for acquiring a sense of justice. Okin’s critique of the family is nevertheless not a radical one. She merely insists that Rawlsian principles of justice be universal in scope and extend to all institutions, including the family. The second line of feminist criticism concerns the alleged maleness of a preoccupa- tion with justice and rights. Appeal is made to a distinctive female ethics which empha- sizes attachment, responsibility, context and particularity as opposed to independence, rights, abstraction and universality. The idea that women speak in another moral ‘voice’ to that of men is due principally to the work of Carol Gilligan (1982) in the psychol- ogy of moral development. But the contrast between an ethics of care and one of justice now enjoys wide currency in moral and political philosophy. Gilligan herself does not see these moral outlooks as mutually exclusive; nor does she see each of them as nec- essarily associated with one particular gender. Indeed she seems to favour an account of moral development which emphasizes the socializing role of parents, whatever their gender. From the standpoint of social and political philosophy this particular line of criti- cism bites only if the prevailing liberal accounts of the good society may be judged defective for omitting mention of care. It begs too many questions to suggest that justice 272 POLITICAL AND SOCIAL PHILOSOPHY applies in the public sphere, and care in the private. It would also grossly over-simplify to suggest that social co-operation could be governed either by rules of justice or by an ethos of care. Indeed there are reasons to agree with the Rawlsian view that justice has primacy. Care alone cannot determine who should be in receipt of what goods. And where care fails or falls short of what is desired, justice speciﬁes what can be legitimately expected of others. Nevertheless, the opposition between care and justice highlights – as does the communitarian critique – the extent to which a society governed solely by respect for rights and a ‘sense of justice’ may not generate any real sense of relatedness	Blackwell
and interdependence among its members. 5.3 Analytical Marxism As the title implies analytical Marxists have displayed the virtues of anglophone philosophy in general, that is argumentative rigour, scrupulous attention to the text, and careful conceptual analysis. Whether such work is Marxist is debatable. It certainly eschews wholesale subscription to every aspect of Marx’s work, and prefers instead separately to appraise the distinct claims that may be said to constitute this theory. It has rejected a certain understanding of history, TELEOLOGICAL (pp. 319–20) and HEGELIAN (chapter 33) in origin, and, in the case of Jon Elster (1985) at least, embraced a robust METHODOLOGICAL INDIVIDUALISM (pp. 397–9) that is very unfamiliar among Marxists. It has been willing to jettison what many would see as central Marxist claims, for instance the LABOUR THEORY OF VALUE (chapter 34), and has increasingly conceded the attractiveness of non-Marxian theories in political philosophy. However, analytical Marxism has insisted upon the speciﬁc character of capitalism, and sought to explicate the manner both in which such a society is fundamentally ﬂawed, and in which socialism represents a feasible and desirable alternative. Yet there are deep problems with such an approach. In the ﬁrst instance it needs to be shown that Marx had a theory of justice by which CAPITALISM (pp. 755–6) can be judged unjust. This task confronts a familiar paradox. Marx employs a language with appar- ent moral import, yet explicitly disdains moral criticism and theory (Lukes 1985). The paradox is not easy to resolve, and any resolution may simply have to accommodate the fact that Marx was not consistent in his outlook (Geras 1985). Second, if capitalism is unjust then for Marxists that injustice must inhere in some signiﬁcant feature of the relationship between capitalist and proletarian. Two candi- dates suggest themselves. The ﬁrst is that the capitalist exploits the proletarian. Yet on any plausible account of exploitation that can be offered it is arguable that some form of exploitation may characterize every society, including socialism. Moreover, the difference in ownership of resources which deﬁnes the capitalist relation need not be unfair. It may have arisen in a perfectly legitimate fashion, for instance through the superior efforts, talents, assiduity and prodigality of the capitalist. The idea of a ‘cleanly generated capitalism’ conforms to Robert Nozick’s prescription of a just distribution that has arisen by just steps from an originally just situation. The second candidate for explaining the injustice of the capitalist relation is coercion. Workers may be unfairly forced to work for their employers. The difﬁculty is that for any one worker the alternatives to employment are not so stark and restrictive as would be required to establish coercion. This is especially true in a modern 273 DAVID ARCHARD welfare society where it must also be acknowledged that avenues of escape from one’s class do exist for the talented and hard working. Difﬁculties in appreciating the speciﬁc inequity of capitalism are compounded by imprecision in the recommendations of socialist justice. Notoriously Marx said little about the lineaments of the future society, and what he did say suggests utopianism in the pejorative sense of this word. Talk of a society ‘beyond justice’ might imply the view that the circumstances of justice will be transcended. This is unrealistic. Or it might imply the irrelevance of any evaluative criteria to such a society. This is dangerously naive. Marxists can engage with other contemporary political philosophers on the grounds of equality, emphasizing the deep, structured inequalities of present capitalist society and the manner in which these inequalities corrupt any formal equality of	Blackwell
civic rights. But if they are to be radical egalitarians they may differ little from some liberals, or be beset by difﬁculties which derive from their commitment to principles, such as that of self-ownership, which liberals do not share (Cohen 1990). There is a ﬁnal important point. Any theory which indicts the present society of fundamental injustice and recommends a future perfect society needs an account of the means of transforming the ﬁrst into the second. Traditionally Marxism relied on some combination of the historical guarantee of revolutionary change and the uniquely important role of the proletariat. Analytical Marxism is not teleologically optimistic and has tended to reticence about proletarian activism. This is partly due to its individualist presuppositions, which make it hard to appreciate the factors disposing to and inhibiting COLLECTIVE AGENCY (pp. 397–9). It is also due to the need for an account of why what is in the interests of the majority of society’s members coincides with the morally desired emancipation of all. This is increasingly difﬁcult, for there is no longer a group of that orthodoxy – being the exploited bulk of capitalist society with nothing to lose and everything to gain by overthrowing that society and its injustices. individuals who satisfy the various requirements of Analytical Marxism has performed a signal service by bringing Marxism within the fold of contemporary political philosophy. But it may have done so at the price of expos- ing serious if not fatal shortcomings in Marxism. And to the extent that analytical Marxists have become political philosophers they have arguably ceased to be Marxists. 6 Individuals and Communities A standard charge against liberalism is that it is individualistic. Methodological indi- vidualism is discussed elsewhere, and the various senses in which liberalism might be said to neglect community were considered in the section on communitarianism. What needs to be examined here is the signiﬁcance political philosophy should accord to groups or communities, and the relationship of individuals to these collective entities. For while each of us is an individual we are also social creatures, belonging to particular tribes, cultures, religions, races and nations. To deny these facts would be productive of an implausible theory and an impoverished political practice. There are at least two ways in which group identity is signiﬁcant. In the ﬁrst instance there is the question of how the state should treat the existence within its jurisdiction of stable, enduring, well-deﬁned groups with their own history, culture and way of life. 274 POLITICAL AND SOCIAL PHILOSOPHY This is the problem of cultural pluralism. In the second instance there is the question of where the boundaries of any state or jurisdiction should be drawn and what role should be played in this context by nationhood. This is the problem of nationality and nationalism. 6.1 Cultural pluralism Although political philosophy has, from its inception with the Greeks, tended to assume the cultural or ethnic homogeneity of the ‘people’ whose obligation a legitimate state commands and to whom a set of principles of justice may apply, the fact is that all modern societies comprise distinct stable groups whose members identify themselves – and are identiﬁed by others – by reference to some combination of shared race, religion, nationality, language, culture or history. How should the state respond? It could insist on denying the fact of difference either by enforcing a ‘republican’ ideal wherein citizens have no allegiances or identities other than those which consti- tute them as members of the polity, or by supporting assimilationist practices whereby members of cultural minorities must acquire the identity of the dominant community. Such measures of compulsory homogenization are widely perceived as unfair in	Blackwell
denying to individuals something of great value, namely the expression of their own particular communal identity. A culturally pluralist state, by contrast, honours the existence of plural identities by measures which may range from underwriting the right of persons belonging to minorities to enjoy their own culture, to ‘communalist’ measures which positively protect and preserve the distinct groups recognized within society. Liberals are unwilling to see the warrant for such measures in the existence of group rights which do not straightforwardly reduce to the rights of the individual members of the group in question. Rather they have argued that the protection of groups can be defended in so far as doing so protects and advances the interests of individuals as members of groups (Raz 1986: 207–9). The value of a culture is said to be that of cul- tural membership, its value to the individuals who are members (Buchanan 1991). In an original and arresting argument, Kymlicka (1989) rejects the idea of group rights but commends a policy of actively seeking to preserve cultures. He does so by arguing that our cultural membership is a good in so far as it provides the necessary context from within which we are able to make and evaluate meaningful, rational, autonomous choices of life. Charles Taylor (1992) has tried to defend a liberalism which might permit the state to nourish and protect a particular culture, so long as it was also able to secure the rights of those who do not share the dominant culture. The major problem is that liberalism may be left with no other criterion for apprais- ing different cultures than the extent to which they nurture the kind of individuals favoured by liberals. Should a liberal society tolerate a minority culture which does not respect autonomy, even if it supplies its members with an otherwise worthwhile life? Is cultural pluralism a good thing only if the various cultures are all consistent with liberal ideals of individuality? Moreover any account of cultural pluralism must acknowledge the possible disadvantages of diversity. These may include the increased possibility of social conﬂict, and the undermining of the shared sense of community which is necessary for good political order. 275 DAVID ARCHARD 6.2 Nationalism and nationality Political philosophy has, until recently, remained largely silent on the questions of nationalism or nationality; or it has dismissed such questions as somehow unworthy of philosophical consideration (Pettit and Goodin 1993: 7). Yet there is no more salient fact about the contemporary world, nor any more potent source of conﬂict and violence, than the existence, actual and disputed, of nations. It is also noteworthy that political philosophy has assumed the existence of distinct nation-states and concerned itself with the application to such entities of principles of justice, equality and rights (Canovan 1996). Although they are closely related terms, often used as synonyms, and frequently hyphenated, ‘state’ and ‘nation’ have distinct meanings. A nation is a community of people bound together over time by some signiﬁcant, shared characteristic such as language, race or culture. A state is an independent, sovereign political association of people inhabiting a bounded territory. Nationalism, as a doctrine, makes the factual claim that humanity is and always has been naturally divided into nations. The normative claims of nationalism are that nations should be states and that states should be nations. Contemporary philosophical defenders of nationalism (Miller 1995; Tamir 1993) have sought to answer the criticisms of claims of nationalism, and to show that a defence of nationalism is consistent with, indeed demanded by, a proper defence of liberalism. To the charge that nations are ﬁctive products of modernity, defenders of nationalism will insist that modern nations do have premodern ethnic origins, and	Blackwell
that even false beliefs can have instrumental value if they sustain a valuable sense of community. States should be nations because a principle of nationality may supply the ‘fellow feeling’ J. S. Mill thought necessary for good government (Mill 1975: ch. 16) or for the acceptance of redistributive principles of justice (Miller 1995). Nations should be states because democratic self-government is coextensive with national self-determination, and because membership of a nation being a constitutive element of individual identity and well-being, statehood is an essential means of protecting nationality. To those who insist that the defence of nationality represents an unwarranted partiality inconsistent with the cosmopolitanism which properly realizes the global scope of any acceptable political philosophy, the friends of nationalism will insist that it is necessary to make an accommodation with the realities of the modern world. Further partiality is not morally unjustiﬁed so long as its demonstration remains constrained by a recognition of the minimum owed to those who are not one’s own co-nationals. However, critics of nationalism will reply that since there is no well-bounded territory containing an ethnically homogeneous population the demands of nationalism are productive of grossly unacceptable political results: at best discrimination against national minorities within the state, at worst the forcible expulsion and slaughter of these minorities. Moreover the number of potential, aspirant nation-states greatly exceeds the number of possible viable states, and internal secession would continue to the point of non-viability. Sympathetic political philosophical treatments of nationality may thus stop short of conceding the full demands of nationalism, arguing, for instance, that those ends 276 POLITICAL AND SOCIAL PHILOSOPHY thought to require the coincidence of nation and state may be served by something less, such as federalism (Buchanan 1991). Unsympathetic treatments of nationality may insist that it is possible to construct a non-national principle of political community, around for instance loyalty to the constitutive principles of a particular polity (Habermas 1992). Both camps may seek to defuse the sting of nationalism by encouraging a political disassociation of ‘nation’ and ‘state’. This may be managed through the development of trans- and international institutions, the protection of subnational ethnic plurality, and the provision for decentralized political representation. Political philosophy cannot now ignore the brute facts of nationality. A proper acknowledgement of these facts should consist in a recognition of what can be changed and what cannot, and, in consequence, of the fact that whatever value national identity has may be secured only at the expense of inseparable disadvantages. 7 Political Philosophy and Politics This last section considers what is political philosophy and what is politics. First, then, what is it thought that a political theory should do? I do not mean by this the question of whether a theory should be about justice, the state, the class struggle, or whatever. I mean what is a theory taken to be doing when it answers a self-chosen question such as ‘What is justice?’ There are four broad sorts of answer. The ﬁrst is that a theory should be built upon FOUNDATIONS (pp. 40–1) that are uni- its procedure of argumentation and versally recognized and evident truths; reasoning should similarly be unexceptionable and acceptable to any rational person. It is likely that these foundations will comprise some understanding of the individual human being. Both Rawls’s A Theory of Justice and Nozick’s Anarchy, State, and Utopia may be rep- resented as political theory in this foundational sense. The problem with this approach is that it risks being empty or contentious. Either the foundational understanding of	Blackwell
the human being commands universal agreement in virtue of being emptied of real substantive content, or is a recognizably interesting view but thereby partial and controversial. A familiar criticism of Rawls is that his contracting individuals are not those of any time and place, but historically and culturally located agents – Western, liberal men. A response to these problems is self-consciously to adopt a stance that is rooted in one’s own culture and history. Borrowing Plato’s famous metaphor, Michael Walzer contrasts a philosophical attitude that walks out of the cave and seeks the high ground of universal objectivity with one that takes its stand in the cave and with its inhabi- tants. This ‘way of doing philosophy is to interpret to one’s fellow citizens the world of meanings that we share’ (Walzer 1983: xiv). The problems with this approach are threefold. First, as has already been noted, it courts relativism: the good society is each and every society that judges itself to be good by its own lights. Second, a society’s practices, however unjust they may seem to the reasonable outsider, are not open to criticism if, on the inside, that society lacks the shared meanings which could inform such a critique. Third, the approach tends to conservatism. It implies that a society’s meanings are deﬁned by its existing practices. But there is no space from which to judge 277 DAVID ARCHARD these practices. Or, put another way, the meanings by which the practices could be judged appear to ﬂoat outside the society (Cohen 1986). Rawls appears to have moved from his original foundationalism towards this second understanding of political theory. In Political Liberalism (Rawls 1993) he views the liberal conception of justice as expressing ideas implicit within the institutions and public culture of constitutional democratic regimes. Yet this is not a simple shift from universal to particular, from transcendental rules of justice to the particular virtues of American constitutionalism. ‘Kantianism in one country’ is a ﬁne and witty deﬁnition of Rawls’s political philosophy. It is possible, and plausible, to believe that modern democracy is an historical achievement which institutes general moral principles – equality of respect for individual choices of life, and the public justiﬁability of agreed rules of social co-operation. Actually existent institutions may do no more than approximate to these ideals, but they nevertheless aspire to them. At the same time Rawls would now insist that political philosophy is turned to when a society’s shared understandings break down and come into conﬂict. Philosophy resolves these conﬂicts by ascending to abstractions which are nevertheless also somehow uncovered through fundamental ideas deep and implicit within the society’s culture (ibid.: 44–6). This is somewhat mysterious. It may also betray a false optimism that the plurality of fundamental beliefs which characterizes modern society is, at some level, eliminable. On the other hand, if the implication is that we can reach outside the terms of our present, public culture for values which validate it, then too many crucial questions have been begged. This at least is the charge of Richard RORTY (pp. 783–4) (1989, 1991). Whether or not democratic liberalism stands at the end of ideological history is a question which anglophone philosophy is on the whole ill-equipped to acknowledge or answer, not least because it lacks the means to theorize ideas of modernity and historical progress. Philosophy in the English-speaking world is notoriously incapable of ‘reﬂexive social understanding’; that is, being in a position to interrogate itself about its relation to the society and history in which it is situated (Williams 1980). One way to open a space between presently shared meanings and preferred alterna-	Blackwell
tives is given by the third approach to political philosophy which may be termed CRITIQUE (chapter 32). Marx famously remarked that science would be unnecessary if reality and appearance coincided. His scientiﬁc concern was to disclose the real workings of capitalism which could not be apparent to those who lived and worked under it. Analogously the task of political philosophy may be to reveal what our shared mean- ings do not and cannot say about our political world. The function of critique is to display these gaps and thus the distance between actuality and the moral pretensions generated by that actuality. To the extent that analytical Marxism shares many of the premises and concepts of contemporary philosophical liberalism, it may be said to have eschewed critique in favour of criticism. However, feminist political theory does aspire to critique when it seeks to show how unspoken assumptions – about women’s nature, the public–private divide, the family – explain why liberalism is both silent about, and yet eloquent in its implicit endorsement of, women’s continued subordination. The fourth and ﬁnal approach to political philosophy is hostile to any misjudged rationalist ambitions theory might have. It repudiates the idea that political association must self-consciously realize any desired end or purposes. Instead it sees the tasks of 278 POLITICAL AND SOCIAL PHILOSOPHY political theory as more modest and restricted. It is the articulation of the commonly acknowledged rules of conduct which inform our actual practices. It salutes the his- torical achievements of modern liberalism, not least the emergence of ‘individuality’ and ‘civility’; that is, law and civic order. Yet it refuses to see these eminently practical achievements as conforming to universal, objective principles that may be laid bare. Political theorizing should be faithful to the knowledge already presupposed in our practice; it should ‘pursue the intimations’ of established traditions. The author of this approach is Michael Oakeshott (1962, 1975). His sceptical and nuanced conservatism does not ﬁgure large in most surveys of contemporary political philosophy, perhaps because he speaks at a tangent to its main concerns. Yet his work has been inﬂuential in a British conservatism, ably represented by Roger Scruton (1984), which keeps its distance from both American philosophical liberalism and the free market, anti-statism that often passes for contemporary right-wing thinking. At the same time there is in such work a Hegelian sensitivity to history and practice which importantly distinguishes it from the simple appeal to shared meanings which characterizes the second approach. Our second important question is, ‘What is politics?’ That is, what is the scope of political activity, what distinguishes it from other forms of human activity, and what is its importance? Here it is best to consider a number of oppositions. First, there is that between what we could call instrumental and expressivist understandings of politics. According to the instrumental understanding, politics is a means used by essentially independent individuals to secure the agreements which are necessary if they are to obtain the beneﬁts of co-operation and avoid the costs of non-agreement. It is the politics of bargaining, compromise, accommodation, and achieving a modus vivendi. The political sphere is like a market-place in which individuals come to make their separate deals and then return to live their lives. Politics has no further function than is necessary to facilitate and protect the agreements made. On the EXPRESSIVIST (pp. 384–90) understanding, political activity is itself valuable. It is an important way in which human beings express themselves as social, co- operative creatures enjoying an interdependent existence. This is the politics of participation, community and republican citizenship. The political sphere is a forum in	Blackwell
which people come together as citizens, and a vital constituent of the full life led by all. Politics is impoverished to the extent that it falls short of exhibiting this character. On the whole liberals, and certainly libertarians, have favoured the former under- standing. Their preference derives from according a primary importance to liberty and believing that individual purposes will be various. Political activism, of the sort envis- aged by expressivists, is unlikely to be spontaneously universal, and will be unaccept- able if coerced. Perhaps in part for this reason liberals, despite a clear preference for it as a regime, have had little to say about how democracy should actually work and what particular form it should take. Democracy tends to be seen as merely the formalized process by which SOCIAL CHOICES (pp. 391–5) are generated from the rational prefer- ences of individuals. Famous paradoxes and difﬁculties characterize the translation of many individual choices into a single public choice. By contrast there is an important and inﬂuential deliberative model of democratic politics. A key ﬁgure is Jürgen Habermas (1992), who has sought to demonstrate the principled presuppositions of everyday communication, such as the speaker’s implicit 279 DAVID ARCHARD claims to be truthful, right and sincere. From these foundations he defends a view of moral norms as those which would, and could, be agreed upon by the members of a communicative community who recognize only the force of the better argument. The relevance of such ideas to an ideal of deliberative or discourse politics whereby individuals reach reasoned agreement on the collective good is evident. The problems are also clear. If the model is intended to be a description of the actual political world it is at best unrealistic, and arguably undervalues disagreement and contestation as constitutive features of political activity. Politics may be ineliminably agonistic. If the model is an ideal then it needs to be shown that it does not import and employ values, such as equality, which cannot be found solely in the nature and presuppositions of conversational speech. The concern of socialists and feminists with democratic theory and practice is attrib- utable in part to the expressivist ideal of political participation as a good (Pateman 1970). It is also due to a worry about reconciling the formal equality of political citi- zenship with the various inequalities, of race, class, and gender, which characterize civil society. Related to but distinct from the instrumental–expressivist distinction is that between public and private, already introduced in section 5. As this distinction is now under- stood, the public comprises the political, legal and economic, the world in which indi- viduals work, vote and are accountable to the rest of society for their actions. The private is the personal and familial, the sphere of the household in which individuals love, play and generally retreat from the public world. To this distinction corresponds differences of motivation, relatedness and loyalty. The public world is cold, impersonal, governed by abstract rules, in which independent and mutually disinterested individuals meet. The private is the emotionally warm haven in which individuals are bound by particular relationships of affection, loyalty and mutual interdependence. Liberals have, on the whole, refused to see the private as political. Indeed the private sphere is that with which the state has no business and into which it should not intrude. This is plausible in so far as liberals have characterized private activity as self- regarding in J. S. Mill’s sense; that is, whose harmful effects, if any, are conﬁned to the agent. Consensual sexual behaviour and procreative decisions, according to liberals, are paradigmatically private in this regard. It is to feminism that we owe a critique of the public–private divide. This insists ﬁrst	Blackwell
that the scope and character of the private is determined publicly by law and policy. It is not that the private domain pre-exists and limits that of the public. Rather the terms of privacy are set within the public sphere. Indeed the very inauguration of the private, and civil society, may be due to an unnoticed but deeply illegitimate ‘sexual contract’ in which the male ﬁrst subordinated the female to serve his sexual purposes (Pateman 1988). Second, the private is political if ‘political’ extends to describe any structure or relationship in which some individuals exercise signiﬁcant power over others. The family is the object of interest here. The traditional institution is marked by a familiar sexual division of labour with the husband dominant over the wife. Any characteriza- tion of familial relations as private amounts to a public endorsement of their continu- ing inequity. 280 POLITICAL AND SOCIAL PHILOSOPHY Even on their own principles liberals should be interested in the family. For it can support or undermine broader social justice. This is not only in so far as the family is, as noted earlier, a site of moral education, a place where a sense of justice is learned. It is also that the family is the major means by which assets – property as well as natural endowments – are transferred across generations. Differences between familial situations remain a crucial obstacle to securing equality of opportunity (Fishkin 1983). The ﬁnal opposition sets the political against the economic. Although the economic is included within the public sphere, there is a tendency among liberals to be relatively indifferent towards various economic arrangements. These, to simplify their view, are to be preferred on grounds of efﬁciency not morality. In A Theory of Justice Rawls leaves open the choice between a private property economy and socialism (Rawls 1972: 258). In Political Liberalism he maintains that the question is ‘not settled at the level of ﬁrst principles of justice’, but depends on the contingencies of a country’s particular institutions and historical circumstances (Rawls 1993: 338). But the economic is political in three important regards. First, liberal principles of justice apply to the distributive sphere, that is they deter- mine who gets what goods once they are produced. But principles restricted to distrib- ution leave out of account the productive sphere, concerning what is produced by whom. Decisions may be taken as to what goods are produced and in what quantities. It can sometimes seem as if the goods to be distributed in liberal theories of justice have just dropped from heaven, quite explicitly in Bruce Ackerman’s (1980) account, where he speaks of manna. Second, there is a stronger claim associated with Marx that production determines distribution; that is, that a distribution of goods to be consumed is a consequence of how production itself is organized. If this determinist thesis is true then principles of distributive justice are in an important sense beside the point. Third, the economic sphere is one to which political considerations directly apply. Work constitutes a signiﬁcant part of an individual’s life. In itself it can be rewarding or frustrating, self-realizing or alienating. It can be self-directed or performed under conditions of subordination to another. Work is also instrumentally valuable in so far as payment for employment determines one’s level of subsistence and the character of one’s work affects one’s social status and self-esteem. It is important then to recognize that employment – an individual’s right to it, and the determination of its conditions – are proper subjects for political inquiry. Socialists have attended to these issues. They have defended an extension of democracy to the work place, and have been more generally concerned about the proper balance between	Blackwell
market and state control, private and public ownership. 8 Conclusion Political philosophy is alive and well. It was not quite moribund in 1950 but the post-1970 revival now makes it appear so. The current dominance of philosophical liberalism may irk those who feel that its approach pre-emptively biases consideration important matters and ignores crucial facts. But those who feel that way must of 281 DAVID ARCHARD recognize that it is to liberals that we owe the revival of political philosophy and our present ability to speak to the problems of our social and political existence. We may not wish to speak in the language of liberalism. But then philosophers can no longer, as Weldon once argued, be silent, and the onus is on those who dissent to perfect and practise the alternative languages of political criticism. Further Reading Of the general surveys of social and political philosophical work, Kymlicka (1990) is the best. It is comprehensive, incisive and judicious. Plant (1991) is well-informed and successfully the main concerns of contemporary political theory (though he brings out many of unaccountably ignores feminism). Pettit (1980) and Brown (1986) are also useful if more narrowly concerned with the main theories of distributive justice. Philosophy and Public Affairs remains the ﬂagship journal of political philosophy in the present era. Goodin and Pettit (1993) is a volume in the present series. It contains extended essays on various disciplinary contributions to the subject, shorter entries on major ideologies, and short notes on special topics. The contributors are distinguished practitioners of the subject, the writing is almost uniformly excellent, and the whole text supplies an illuminating picture of post-Rawlsian political philosophy. The Oxford Political Theory series, edited by David Miller and Alan Ryan, contains excellent books by distinguished and particularly well qualiﬁed contributors on particular topics within contemporary political philosophy. On Rawls the literature is voluminous. Daniels (1975) collects the best of the earlier critical essays, and other standard commentaries are Barry (1973) and Wolff (1977). Kukathas and Pettit (1990) compares ‘early’ and ‘late’ Rawls, but before publication of Political Liberalism. Paul (1981) does for Nozick what Daniels does for Rawls, and Wolff (1991) offers an unprejudiced but critical account of Nozick’s theory. Mulhall and Swift (1996) provides a ﬁne, clearly written review of the debate between it struggles somewhat to ﬁnd the common liberals and their communitarian critics, even if themes in communitarianism. In addition to Gutman (1985), the following also supply useful evaluations of communitarianism: Kymlicka (1988) and Buchanan (1989). G. A. Cohen has worried at and about the legacy of Marx and its relationship to current political philosophy with more acuity and assiduity than anyone. Cohen (1988) collects together some of his best pieces. Kittay and Meyers (1987) gives a good sense of Gilligan’s impact upon moral and political thinking. Jaggar (1983) is a standard introduction to feminist political theory, while Grimshaw (1986) is also clearly written and relevant. An introduction to British conservatism is provided by Covell (1986), and Franco (1990) offers a comprehensive, if cautious account of his subject. Kymlicka (1995) is a comprehensive collection of important articles on cultural and national pluralism, while Beiner (1999) offers a ﬁne collection of pieces by leading theorists on the topic of nationalism. Chambers (1996) is a very good evaluation of the ideal of Habermasian deliberative democracy within the context of Rawlsian political philosophy. References Ackerman, B. 1980: Social Justice in the Liberal State. New Haven, CT: Yale University Press. Barry, B. 1973: The Liberal Theory of Justice. Oxford: Clarendon Press. 282	Blackwell
POLITICAL AND SOCIAL PHILOSOPHY Beiner, R. (ed.) 1999: Theorizing Nationalism. Albany: State University of New York Press. Berlin, I. 1958: Two Concepts of Liberty. In Four Essays on Liberty. Oxford: Clarendon Press. Brown, A. 1986: Modern Political Philosophy. Harmondsworth: Penguin Books. Buchanan, A. 1989: Assessing the Communitarian Critique of Liberalism. Ethics, 99 (July), 852–82. —— 1991: Secession: The Morality of Political Divorce from Fort Sumter to Lithuania and Quebec. Boulder, CO: Westview Press. Caney, S. 1992: Liberalism and Communitarianism: A Misconceived Debate. Political Studies, 40, 273–89. Canovan, M. 1996: Nationhood and Political Theory. Cheltenham: Edward Elgar. Chambers, S. 1996: Reasonable Democracy: Jürgen Habermas and the Politics of Democracy. Ithaca, NY: Cornell University Press. Cohen, G. A. 1978: Karl Marx’s Theory of History: A Defence. Oxford: Oxford University Press. —— 1988: History, Labour, and Freedom: Themes from Marx. Oxford: Clarendon Press. —— 1990: Marxism and Contemporary Political Philosophy, or: Why Nozick Exercises some Marxists more than he does any Egalitarian Liberals. Canadian Journal of Philosophy, supple- mentary volume 16, 363–87. Cohen, J. 1986: Review of Walzer, Spheres of Justice. Joumal of Philosophy, 83: 8, 457–68. Covell, C. 1986: The Redeﬁnition of Conservatism. Basingstoke: Macmillan. Daniels, N. (ed.) 1975: Reading Rawls. Oxford: Blackwell. Dworkin, R. 1978: Liberalism. In S. Hampshire (ed.) Public and Private Morality. Cambridge: Cambridge University Press. —— 1981: What is Equality? Part 1: Equality of Welfare and Part 2: Equality of Resources. Philosophy and Public Affairs, 10: 3, 185–246, and 10: 4, 283–345. Elster, J. 1985: Making Sense of Marx. Cambridge: Cambridge University Press. Fishkin, J. S. 1983: Justice, Equal Opportunity, and the Family. New Haven, CT: Yale University Press. Franco, P. 1990: The Political Philosophy of Michael Oakeshott. New Haven, CT: Yale University Press. Galston, W. A. 1980: Justice and the Human Good. Chicago: University of Chicago Press. Geras, N. 1985: The Controversy about Marx and Justice. New Left Review, 150, 47–85. Gilligan, C. 1982: In a Different Voice. Cambridge, MA: Harvard University Press. Goodin, R. E. and Pettit, P. (eds) 1993: A Companion to Contemporary Political Philosophy. Oxford: Blackwell. Gray, J. 1984: Hayek on Liberty. Oxford: Blackwell. Grimshaw, J. 1986: Feminist Philosophers. Brighton: Wheatsheaf. Gutman, A. 1985: Communitarian Critics of Liberalism. Philosophy and Public Affairs, 14: 3, 308–22. Habermas, J. 1992: Citizenship and National Identity: Some Reﬂections on the Future of Europe. Praxis International, 12, 1–33. —— 1996: Moral Consciousness and Communicative Action (translated by Christian Lenhardt and Shierry Weber Nicholsen). Cambridge, MA: MIT Press. Haksar, V. 1979: Equality, Liberty and Perfectionism. Oxford: Oxford University Press. Jaggar, A. M. 1983: Feminist Politics and Human Nature. Totowa, NJ: Rowman and Allanheld. Kittay, E. F. and Meyers, D. T. (eds) 1987: Women and Moral Theory. Savage, MD: Rowman and Littleﬁeld. Kocis, R. A. 1980: Reason, Development, and the Conﬂict of Human Ends: Sir Isaiah Berlin’s Vision of Politics. American Political Science Review, 74: 1, 38–52. Kukathas, C. and Pettit, P. 1990: Rawls: A Theory of Justice and its Critics. Cambridge: Polity Press. 283 DAVID ARCHARD Kymlicka, W. 1988: Liberalism and Communitarianism. Canadian Journal of Philosophy, 18, 2 (June), 181–204. —— 1989: Liberalism, Community, and Culture. Oxford: Clarendon Press. —— 1990: Contemporary Political Philosophy. Oxford: Clarendon Press. —— (ed.) 1995: The Rights of Minority Cultures. Oxford: Oxford University Press. Landesman, B. 1983: Egalitarianism. Canadian Journal of Philosophy, 13, 27–56. Laslett, P. 1950: Introduction. In P. Laslett (ed.) Philosophy, Politics and Society. Oxford: Blackwell. Locke, J. 1689: Epistola de Tolerantia. English translation, A Letter on Toleration (introduction and	Blackwell
notes by J. W. Gough). Oxford: Clarendon Press. Lukes, S. 1985: Marxism and Morality. Oxford: Clarendon Press. MacIntyre, A. 1981: After Virtue. London: Duckworth. —— 1988: Whose Justice? Whose Rationality? London: Duckworth. Mill, J. S. 1975 [1861]: Considerations on Representative Government. In Three Essays. Oxford: Oxford University Press. Miller, D. 1995: On Nationality. Oxford: Clarendon Press. Mulhall, S. and Swift, A. 1996: Liberals and Communitarians, 2nd edn. Oxford: Blackwell. Nagel, T. 1991: Equality and Partiality. Oxford: Oxford University Press. Nozick, R. 1974: Anarchy, State and Utopia. Oxford: Blackwell. Oakeshott, M. 1962: Rationalism in Politics. London: Methuen. —— 1975: On Human Conduct. Oxford: Clarendon Press. Okin, S. M. 1989: Justice, Gender, and the Family. New York: Basic Books. Pateman, C. 1970: Participation and Democratic Theory. Cambridge: Cambridge University Press. —— 1988: The Sexual Contract. Cambridge: Polity Press. Paul, J. (ed.) 1982: Reading Nozick. Oxford: Blackwell. Pettit, P. 1980: Judging Justice. London: Routledge and Kegan Paul. Pettit, P. and Goodin, R. (eds) 1993: A Companion to Contemporary Political Philosophy. Oxford: Blackwell. Plant, R. 1991: Modern Political Thought. Oxford: Blackwell. Popper, K. 1945: The Open Society and its Enemies, 2 vols. Vol. 1: Plato. Vol. 2: Hegel and Marx. London: Routledge and Kegan Paul. Rawls, J. 1972: A Theory of Justice. Oxford: Oxford University Press. —— 1993: Political Liberalism. New York: Columbia University Press. Raz, J. 1986: The Morality of Freedom. Oxford: Clarendon Press. Rorty, R. 1989: Contingency, Irony and Solidarity. Cambridge: Cambridge University Press. —— 1991: Objectivity, Relativism and Truth. Cambridge: Cambridge University Press. Sandel, M. 1982: Liberalism and the Limits of Justice. Cambridge: Cambridge University Press. Scruton, R. 1984: The Meaning of Conservatism, 2nd edn. Basingstoke: Macmillan. Sen, A. 1992: Inequality Re-examined. Oxford: Clarendon Press. Tamir, Y. 1993: Liberal Nationalism. Princeton, NJ: Princeton University Press. Taylor, C. 1985: Philosophy and the Human Sciences: Philosophical Papers, Vol. 2. Cambridge: Cambridge University Press. —— 1989: Cross-Purposes: The Liberal Communitarian Debate. In N. Rosenblum (ed.) Liberalism and the Moral Life. Cambridge, MA: Harvard University Press. —— 1992: Multiculturalism and The Politics of Recognition (commentary by A. Gutman, edited by S. C. Rockefeller, M. Walzer and S. Wolf). Princeton, NJ: Princeton University Press. Unger, R. M. 1975: Knowledge and Politics. New York: Free Press. Walzer, M. 1983: Spheres of Justice. New York: Basic Books. Weldon, T. D. 1953: The Vocabulary of Politics. Harmondsworth: Penguin Books. Williams, B. 1980: Political Philosophy and the Analytical Tradition. In M. Richter (ed.) Political Theory and Political Education. Princeton, NJ: Princeton University Press. 284 Wolff, J. 1991: Robert Nozick. Cambridge: Polity Press. Wolff, R. P. 1977: Understanding Rawls. Princeton, NJ: Princeton University Press. POLITICAL AND SOCIAL PHILOSOPHY Discussion Questions 1 Are equality and liberty incompatible ideals? 2 Is value pluralism an ineliminable feature of modern society? 3 Are there any feasible ways in which the circumstances of justice could be transcended? 4 How egalitarian is the ‘difference principle’? 5 What roles does the contractarian argument play in Rawls’s theory of justice? 6 Should a theory of justice be ‘metaphysical’ or ‘political’? 7 Is the distribution of natural assets morally arbitrary? 8 Are rights ‘side constraints’ on actions? 9 Can we be said to own ourselves? 10 Is justice a matter of legitimate acquisition of goods, whatever the pattern of their distribution? 11 What should there be equality of? 12 Should autonomy rather than liberty be the central value of liberalism? 13 Is the idea of a ‘liberal community’ a contradiction in terms? 14 Does liberalism depend on an unacceptable conception of the individual? 15 Is a caring society morally preferable to a just one?	Blackwell
16 Can the family ever be a force for justice? 17 What for Marxists is wrong with capitalism? 18 Is ‘analytical Marxism’ Marxism at all? 19 Can political philosophy go beyond what we happen here and now to believe and value? 20 Is the human being a political animal? 21 Is anything really private in the sense of being beyond the public gaze and public concern? 22 Is the ideal of ofﬁcial neutrality on the question of the good life either feasible or desirable? 23 Why should minority cultures be protected or preserved? 24 Does justice lie beyond, between or within nations? 25 What inequalities in our material circumstances are permissible? 26 Is liberal nationalism an oxymoron? 27 Are we all liberals now? 285 9 Philosophy of Science DAV I D PA P I N E AU The philosophy of science can usefully be divided into two broad areas. On the one hand is the epistemology of science, which deals with issues relating to the justiﬁcation of claims to scientiﬁc knowledge. Philosophers working in this area investigate such ques- tions as whether science ever uncovers permanent truths, whether objective decisions between competing theories are possible and whether the results of experiment are clouded by prior theoretical expectations. On the other hand are topics in the meta- physics of science, topics relating to philosophically puzzling features of the natural world described by science. Here philosophers ask such questions as whether all events are determined by prior causes, whether everything can be reduced to physics and whether there are purposes in nature. You can think of the difference between the epis- temologists and the metaphysicians of science in this way. The epistemologists wonder whether we should believe what the scientists tell us. The metaphysicians worry about what the world is like, if the scientists are right. Readers will wish to consult chapters on EPISTEMOLOGY (chapter 1), METAPHYSICS (chapter 2), PHILOSOPHY OF MATHEMAT- ICS (chapter 11), PHILOSOPHY OF SOCIAL SCIENCE (chapter 12) and PRAGMATISM (chapter 36). 1 The Epistemology of Science 1.1 The problem of induction Much recent work in the epistemology of science is a response to the problem of induction. Induction is the process whereby scientists decide, on the basis of various observations or experiments, that some theory is true. At its simplest, chemists may note, say, that on a number of occasions samples of sodium heated on a Bunsen burner have glowed bright orange, and on this basis conclude that in general all heated sodium will glow bright orange. In more complicated cases, scientists may move from the results of a series of complex experiments to the conclusion that some funda- mental physical principle is true. What all such inductive inferences have in common, PHILOSOPHY OF SCIENCE however, is that they start with particular premises about a ﬁnite number of past obser- vations, yet end up with a general conclusion about how nature will always behave. And this is where the problem lies. For it is unclear how any ﬁnite amount of infor- mation about what has happened in the past can guarantee that a natural pattern will continue for all time. After all, what rules out the possibility that the course of nature may change, and that the patterns we have observed so far turn out to be a poor guide to the future? Even if all heated sodium has glowed orange up till now, who is to say it will not start glowing blue sometime in the next century? In this respect induction contrasts with deduction. In deductive inferences the premises guarantee the conclusion. For example, if you know that Either this substance is sodium or it is potassium, and then learn further that It is not sodium, you can conclude with certainty that It is potassium. The truth of the premises leaves no room for the conclusion to be anything but true. But in an inductive	Blackwell
inference this does not hold. To take the simplest case, if you are told, for properties A and B, that Each of the As observed so far has been B, this does not guarantee that All As, including future ones, are Bs. It is perfectly possible that the former claim may be true, but the latter false. The problem of induction seems to pose a threat to all scientiﬁc knowledge. All scientiﬁc discoveries worth their name are in the form of general principles. Galileo’s law of free fall says that ‘All bodies fall with constant acceleration’; Newton’s law of gravitation says that ‘All bodies attract each other in proportion to their masses and in inverse proportion to the square of the distance between them’; Avogadro’s law says that ‘All gases at the same temperature and pressure contain the same number of molecules per unit volume’; and so on. The problem of induction calls the authority of all these laws into question. For if our evidence is simply that these laws have worked so far, then how can we be sure that they will not be disproved by future occurrences? 1.2 Popper’s falsiﬁcationism One inﬂuential response to the problem of induction is due to Sir Karl Popper (1902–94). In Popper’s (1959a, 1963, 1972) view, science does not rest on induction in the ﬁrst place. Popper denies that scientists start with observations, and then infer a general theory. Rather, they ﬁrst put forward a theory, as an initially uncorroborated conjecture, and then compare its predictions with observations to see whether it stands up to test. If such tests prove negative, then the theory is experimentally falsiﬁed, and the scientists will seek some new alternative. If, on the other hand, the tests ﬁt the theory, then scientists will continue to uphold it – not as proven truth, admittedly, but nevertheless as an undefeated conjecture. If we look at science in this way, argues Popper, then we see that it does not need induc- tion. According to Popper, the inferences which matter to science are refutations, which take some failed prediction as the premise, and conclude that the theory behind that pre- diction is false. These inferences are not inductive, but deductive. We see that some A is not-B, and conclude that it is not the case that All As are Bs. There is no room here for the premise to be true and the conclusion false. If we discover that some body falls with 287 DAVID PAPINEAU increasing acceleration (say because it falls from a great height, and so is subject to a greater gravitational force as it nears the earth), then we know for sure that all bodies do not fall with constant acceleration. The point here is that it is much easier to disprove theories than to prove them. A single contrary example sufﬁces for a conclusive disproof, but no number of supporting examples will constitute a conclusive proof. So, according to Popper, science is a sequence of conjectures and refutations. Scien- tiﬁc theories are put forward as hypotheses, and they are replaced by new hypotheses when they are falsiﬁed. However, if scientiﬁc theories are always conjectural in this way, then what makes science better than astrology, or spirit worship, or any other form of unwarranted superstition? A non-Popperian would answer this question by saying that real science proves its claims on the basis of observational evidence, whereas supersti- tion is nothing but guesswork. But on Popper’s account, even scientiﬁc theories are guesswork – for they cannot be proved by the observations, but are themselves merely undefeated conjectures. Popper calls this the ‘problem of demarcation’: what is the difference between science and other forms of belief? His answer is that science, unlike superstition, is at least falsiﬁable, even if it is not provable (Popper 1959a: ch. 2). Scientiﬁc theories are framed in precise terms, and so issue in deﬁnite predictions. For example, Newton’s laws tell us exactly where certain planets will appear at certain times. And this	Blackwell
means that if such predictions fail, we can be sure that the theory behind them is false. By contrast, belief systems like astrology are irredeemably vague, in a way which prevents their ever being shown deﬁnitely wrong. Astrology may predict that Scorpios will prosper in their personal relationships on Thursdays, but when faced with a Scorpio whose spouse walks out on a Thursday, defenders of astrology are likely to respond that the end of the marriage was probably for the best, all things considered. Because of this, nothing will ever force astrologists to admit their theory is wrong. The theory is phrased in such imprecise terms that no actual observations can possibly falsify it. Popper himself uses the criterion of falsiﬁability to distinguish genuine science, not just from traditional belief systems like astrology and spirit worship, but also from Marxism, psychoanalysis and various other modern disciplines that he denigrates as ‘pseudo-sciences’. According to Popper, the central claims of these theories are as unfal- siﬁable as those of astrology. Marxists predict that proletarian revolutions will be suc- cessful whenever capitalist regimes have been sufﬁciently weakened by their internal contradictions. But when faced with unsuccessful proletarian revolutions, they simply respond that the contradictions in those particular capitalist regimes have not yet weak- ened them sufﬁciently. Similarly, psychoanalytic theorists will claim that all adult neu- roses are due to childhood traumas, but when faced by troubled adults with apparently undisturbed childhoods, they will say that those adults must nevertheless have under- gone private psychological traumas when young. For Popper, such ploys are the antithesis of scientiﬁc seriousness. Genuine scientists will say beforehand what obser- vational discoveries would make them change their minds, and will abandon their theories if these discoveries are made. But Marxists and psychoanalytic theorists frame their theories in such a way, argues Popper, that no possible observations need ever make them adjust their thinking. 288 PHILOSOPHY OF SCIENCE 1.3 The failings of falsiﬁcationism At ﬁrst sight Popper seems to offer an extremely attractive account of science. He explains its superiority over other forms of belief, while at the same time apparently freeing it from any problematic dependence on induction. Certainly his writings have struck a chord within the scientiﬁc community. Popper is one of the few philosophers ever to have become a Fellow of the Royal Society, an honour usually reserved for eminent scientists. In the philosophical world, however, Popper’s views are more controversial. This is because many philosophers feel that his account of science signally fails to solve the problem with which he begins, namely, the problem of induction (for example, see Ayer 1956: 71–5; Worrall 1989). The central objection to his position is that it only accounts for negative scientiﬁc knowledge, as opposed to positive knowledge. Popper points out that a single counter-example can show us that a scientiﬁc theory is wrong. But he says nothing about what can show us that a scientiﬁc theory is right. Yet it is positive knowl- edge of this latter kind that makes science important. We can cure diseases and send people to the moon because we know that certain causes do always have certain results, not because we know that they do not. Useful scientiﬁc knowledge comes in the form ‘All As are Bs’, not ‘It’s false that all As are Bs’. Since Popper only accounts for the latter kind of knowledge,heseemsto leave out what is most interesting and important about science. Popper’s usual answer to this objection is that he is concerned with the logic of pure scientiﬁc research, not with practical questions about technological applications. Sci- entiﬁc research requires only that we formulate falsiﬁable conjectures, and reject them	Blackwell
if we discover counter-examples. The further question of whether technologists should believe those conjectures, and rely on their predictions when, say, they administer some drug or build a dam, Popper regards as an essentially practical issue, and as such not part of the analysis of rational scientiﬁc practice. But this will not do. After all, Popper claims to have solved the problem of induction. But the problem of induction is essentially the problem of how we can base judgements about the future on evidence about the past. In insisting that scientiﬁc theories are just conjectures, and that therefore we have no rational basis for believing their predictions, Popper is simply denying that we can make rational judgements about the future. Consider these two predictions: (1) when I jump from this tenth-ﬂoor window I shall crash painfully into the ground; (2) when I jump from the window I will ﬂoat like a feather to a gentle landing. Intuitively, it is more rational to believe (1), which assumes that the future will be like the past, than (2), which does not. But Popper, since he rejects induction, is committed to the view that past evidence does not make any beliefs about the future more rational than any others, and therefore that believing (2) is no less rational than believing (1). Something has gone wrong. Of course believing (1) is more rational than believing (2). In saying this, I do not want to deny that there is a problem of induction. Indeed it is precisely because believing (1) is more rational than believing (2) that induction is problematic. Everybody, Popper aside, can see that believing (1) is more rational than believing (2). The problem is then to explain why believing (1) is more rational than believing (2), in the face of the apparent invalidity of induction. So Popper’s denial of 289 DAVID PAPINEAU the rational superiority of (1) over (2) is not so much a solution to the problem of induc- tion, but simply a refusal to recognize the problem in the ﬁrst place. Even if it fails to deal with induction, Popper’s philosophy of science does have some strengths as a description of pure scientiﬁc research. For it is certainly true that many scientiﬁc theories start life as conjectures, in just the way Popper describes. When Einstein’s general theory of relativity was ﬁrst proposed, for example, very few scien- tists actually believed it. Instead they regarded it as an interesting hypothesis, and were curious to see whether it was true. At this initial stage of a theory’s life, Popper’s recommendations make eminent sense. Obviously, if you are curious to see whether a theory is true, the next step is to put it to the observational test. And for this purpose it is important that the theory is framed in precise enough terms for scientists to work out what it implies about the observable world – that is, in precise enough terms for it to be falsiﬁable. And of course if the new theory does get falsiﬁed, then scientists will reject it and seek some alternative, whereas if its predictions are borne out, then scientists will continue to investigate it. Where Popper’s philosophy of science goes wrong, however, is in holding that sci- entiﬁc theories never progress beyond the level of conjecture. As I have just suggested, theories are often mere conjectures when they are ﬁrst put forward, and they may remain conjectures as the initial evidence ﬁrst comes in. But in many cases the accu- mulation of evidence in favour of a theory will move it beyond the status of conjecture to that of established truth. The general theory of relativity started life as a conjecture, and many scientists still regarded it as hypothetical even after Sir Arthur Eddington’s famous initial observations in 1919 of light apparently bending near the sun. But by now this initial evidence has been supplemented with evidence in the form of gravita- tional red-shifts, time-dilation and black holes, and it would be an eccentric scientist	Blackwell
who nowadays regarded the general theory as less than ﬁrmly established. Such examples can be multiplied. The heliocentric theory of the solar system, the theory of evolution by natural selection and the theory of continental drift all started life as intriguing conjectures, with little evidence to favour them over their competitors. But in the period since they were ﬁrst proposed these theories have all accumulated a great wealth of supporting evidence. It is only those philosophers who have been bemused by the problem of induction who view these theories as being no better than initial hypotheses. Everybody else who is acquainted with the evidence has no doubt that these theories are proven truths. 1.4 Bayesianism If we insist, against Popper, that we are fully entitled to believe at least some scientiﬁc theories on the basis of past evidence, then we are committed to ﬁnding some solution to the problem of induction. One currently popular account of the legitimacy of induc- tion is found within Bayesianism, named after Thomas Bayes (c.1701–61) (Horwich 1982; Howson and Urbach 1989). Bayesians are philosophers who hold that our beliefs, including our beliefs in scien- tiﬁc theories, come in degrees. Thus, for example, I can believe to degree 0.5 that it will rain today, in the sense that I think there is a 50 per cent likelihood of rain today. Simi- larly, I might attach a 0.1 degree of belief to the theory that the strong nuclear and 290 PHILOSOPHY OF SCIENCE electro-weak forces are the same force – I think it unlikely, but allow that there is a one- in-ten possibility it may turn out true. As these examples indicate, Bayesians think of degrees of belief as the extent to which you subjectively take something to be PROBABLE (pp. 167–8). Accordingly, they argue that your degrees of belief ought to satisfy the axioms of the probability calcu- lus. (See the box below for the Dutch Book Argument for this thesis.) It is important to realize, however, that while Bayesians think of degrees of belief as probabilities in this mathematical sense, they still think of them as subjective probabilities. In particular, they allow that it can be perfectly rational for different people to attach different sub- jective probabilities to the same proposition – you can believe that it will rain today to degree 0.2, while I believe this to degree 0.5. What rationality does require, according to the Bayesians, is only that if you have a subjective probability of 0.2 for rain, then you must have one of 0.8 for its not raining, while if I have 0.5 for rain, then I must have 0.5 for its not raining. That is, both of us must accord, in our different ways, with the theorem of the probability calculus that Prob(p) = 1 - Prob(not-p). At ﬁrst sight, this element of subjectivity might seem to disqualify Bayesianism as a possible basis for scientiﬁc rationality. If we are all free to attach whatever degrees of belief we like to scientiﬁc theories, provided only that we are faithful to the structure of the probability calculus, then what is to stop each of us from supporting different theo- ries, depending only on individual fads or prejudices? But Bayesians have an answer; namely, that it does not matter what prejudices you start with, as long as you revise your degrees of belief in a rational way. Bayesians derive their account of how to revise degrees of belief, as well as their name, from Bayes’s theorem, originally proved by Thomas Bayes in a paper published in 1763. Bayes’s theorem states: Prob(H/E) = Prob(H) ¥ Prob(E/H)/Prob(E). The simple proof of this theorem is given in the box. But the philosophical signiﬁcance of the theorem is that it suggests a certain procedure for revising your degrees of belief in response to new evidence. Suppose that H is some hypothesis, and E is some newly discovered evidence. Then Bayesians argue that, when you discover E, you should adjust your degree of belief in H in line with the right-hand side of the above equation: that	Blackwell
is, you should increase it to the extent that you think E is likely given H, but unlikely otherwise. In other words, if E is in itself very surprising (like light bending in the vicin- ity of the sun) but at the same time just what you would expect given your theory H (the general theory of relativity), then E should make you increase your degree of belief in H a great deal. On the other hand, if E is no more likely given H than it would be on any other theory, then observing E provides no extra support for H. The movement of the tides, for example, is no great argument for general relativity, even though it is predicted by it, since it is also predicted by the alternative Newtonian theory of gravitation. Note in particular that this strategy for updating degrees of belief in response to evi- dence can be applied to inductive inferences. Consider the special case where H is some universal generalization – all bodies fall with constant acceleration, say – and the evi- dence E is that some particular falling body has been observed to accelerate constantly. 291 DAVID PAPINEAU If this observation was something you did not expect at all, then Bayesianism tells you that you should increase your degree of belief in Galileo’s law signiﬁcantly, for it is just what Galileo’s law predicts. Of course, once you have seen a number of such observa- tions, and become reasonably convinced of Galileo’s law, then you will cease to ﬁnd new instances surprising, and to that extent will cease to increase your degree of belief in the law. But that is as it should be. Once you are reasonably convinced of a law, then there is indeed little point in gathering further supporting instances, and so it is to the credit of Bayesianism that it explains this. The Bayesian account of how to revise degrees of belief seems to make good sense. In addition, it promises a solution to the problem of induction, since it implies that positive instances give us reason to believe scientiﬁc generalizations. There are, however, problems facing this account. For a start, a number of philoso- phers have queried whether Bayes’s theorem, which after all is little more than an arith- metical truth, can constrain what degrees of belief we adopt in the future (see the box below). And even if we put this relatively technical issue to one side, it is unclear how far the Bayesian account really answers the worry raised above, that the subjectivity of degrees of belief will allow different scientists to commit themselves arbitrarily to dif- ferent theories. The Bayesian answer to this worry was that Bayes’s theorem will at least constrain these different scientists to revise their degrees of belief in response to the evi- dence in similar ways. But, even so, it still seems possible that the scientists will remain on different tracks, if they start at different places. If two scientists are free to attach different prior degrees of belief in Galileo’s law, and both update those degrees of belief according to Bayes’s theorem when they learn the evidence, will they not still end up with different posterior degrees of beliefs? The standard answer to the objection is to appeal to convergence of opinion. The idea is that, given enough evidence, everybody will eventually end up in the same place, even if they have different starting-points. There are a number of theorems of probability theory showing that, within limits, differences in initial probabilities will be ‘washed out’, in the sense that sufﬁcient evidence and Bayesian updating will lead to effectively identical ﬁnal degrees of belief. So in the end, argue Bayesians, it does not matter if you start with a high or low degree of belief in Galileo’s law – for after 1,000 observations of constantly falling bodies you will end up believing it to a degree close to 1 anyway. However, interesting as these results are, they do not satisfactorily answer the fun- damental philosophical questions about inductive reasoning. For they do not work for	Blackwell
all possible initial degrees of belief. Rather, they assume that the scientists at issue, while differing among themselves, all draw their initial degrees of belief from a certain range. While this range includes all the initial degrees of belief that seem at all intuitively plau- sible, there are nevertheless other possible initial degrees of belief that are consistent with the axioms of probability, but which will not lead to eventual convergence. So, for example, the Bayesians do not in fact explain what is wrong with people who never end up believing Galileo’s law because they are always convinced that the course of nature is going to change tomorrow. Of course, Bayesians are right to regard such people as irrational. But they do not explain why they are irrational. So they fail to show why all thinkers must end up with the same attitude to scientiﬁc theories. And in particular they fail to solve the problem of induction, since they do not show why all rational thinkers must expect the future to be like the past. 292 PHILOSOPHY OF SCIENCE Bayesianism The Dutch Book Argument The axioms of probability require that (1) 0 £ Prob(P) £ 1, for any proposition P (2) Prob(P) = 1, if P is a necessary truth (3) Prob(P) = 0, if P is impossible (4) Prob(P or Q) = Prob(P) + Prob(Q), if P and Q are mutually exclusive. Bayesians appeal to the Dutch Book Argument to show why subjective degrees of belief should conform to these axioms. Imagine that your degrees of belief did not so conform. You believe proposition P to degree y, say, and yet do not believe not-P to degree 1 - y. (You thus violate the conjunction of axioms (2) and (4), because P or not-P is a neces- sary truth.) Then it will be possible for somebody to induce you to make bets on P and not-P in such a way that you will lose whatever happens. A set of bets that guarantee that you will lose whatever happens is called a ‘Dutch book’. The undesirability of such a set of bets thus provides an argument that any rational person’s subjective degrees of belief should satisfy the axioms of the probability calculus. Bayes’s Theorem The conditional probability of P given Q - Prob(P/Q) – is deﬁned as Prob(P and Q)/Prob(Q). Intuitively, Prob(P/Q) signiﬁes the probability of P on the assumption that Q is true. It immediately follows from this deﬁnition that Prob(H/E) = Prob(H) ¥ Prob(E/H)/Prob(E) This is Bayes’s theorem. As you can see, it says that the conditional probability Prob(H/E) of some hypothesis H given evidence E is greater than Prob(H) to the extent that E is improbable in itself, but probable given H. Bayesian Updating Bayesians recommend that if you observe some evidence E, then you should revise your degree of belief in H, and set your new Probt’(H) equal to your previous conditional degree of belief in H given E, Probt(H/E), where t is the time before you learn E, and t¢ after. Bayes’s theorem, applied to your subjective probabilities at t, then indicates that this will increase your degree of belief in H to the extent that you previously thought E to be subjectively improbable in itself, but subjectively probable given H. This Bayesian recommendation, that you revise your degree of belief in H by setting it equal to your old conditional degree of belief in H given E, should be distinguished from Bayes’s theorem. Bayes’s theorem is a trivial consequence of the deﬁnition of conditional probability, and constrains your degrees of belief at a given time. The Bayesian recom- mendation, by contrast, speciﬁes how your degrees of belief should change over time. Bayes’s theorem is uncontentious, but it is a matter of active controversy whether there is any satisfactory way of defending the Bayesian recommendation (Hacking 1967; Teller 1973). 293 DAVID PAPINEAU 1.5 Instrumentalism versus realism At this stage let us leave the problem of induction for a while and turn to a different difﬁculty facing scientiﬁc knowledge. Much of science consists of claims about unob-	Blackwell
servable entities like viruses, radio waves, electrons and quarks. But if these entities are unobservable, how are scientists supposed to have found out about them? If they cannot see or touch them, does it not follow that their claims about them are at best speculative guesses, rather than ﬁrm knowledge? It is worth distinguishing this problem of unobservability from the problem of induc- tion. Both problems can be viewed as difﬁculties facing theoretical knowledge in science. But whereas the problem of induction arises because scientiﬁc theories make general claims, the problem of unobservability is due to our lack of sensory access to the subject matter of many scientiﬁc theories. (So the problem of induction arises for general claims even if they are not about unobservables, such as ‘All sodium burns bright orange’. Conversely, the problem of unobservability arises for claims about unobserv- ables even if they are not general, such as ‘One free electron is attached to this oil drop’. In this section and the next, however, it will be convenient to use the term ‘theory’ speciﬁcally for claims about unobservables, rather than for general claims of any kind.) There are two general lines of response to the problem of unobservability. On the one hand are realists, who think that the problem can be solved. Realists argue that the observable facts provide good indirect evidence for the existence of unobservable en- tities, and so conclude that scientiﬁc theories can be regarded as accurate descriptions of the unobservable world. On the other hand are instrumentalists, who hold that we are in no position to make ﬁrm judgements about imperceptible mechanisms. Instru- mentalists allow that theories about such mechanisms may be useful ‘instruments’ for simplifying our calculations and generating predictions. But they argue that these the- ories are no more true descriptions of the world than the ‘theory’ that all the matter in a stone is concentrated at its centre of mass (which is also an extremely useful assump- tion for doing certain calculations, but clearly false). Earlier this century instrumentalists used to argue that we should not even interpret theoretical claims literally, on the grounds that we cannot so much as meaningfully talk about entities we have never directly experienced. But nowadays this kind of semantic instrumentalism is out of favour. Contemporary instrumentalists allow that scientists can meaningfully postulate, say, that matter is made of tiny atoms containing nuclei orbited by electrons. But they then take a SCEPTICAL (pp. 45–56) attitude to such postulates, saying that we have no entitlement to believe them (as opposed to using them as an instrument for calculations). An initial line of argument open to realism is to identify some feature of scientiﬁc practice and then argue that instrumentalism is unable to account for it. One aspect of scientiﬁc practice invoked in this connection has been the uniﬁcation of different kinds of theories in pursuit of a single ‘theory of everything’ (Friedman 1984); other features of science appealed to by realists have included the use of theories to explain observable phenomena (Boyd 1980), and the reliance on theories to make novel predictions (Smart 1963). For, so the realist argues, these aspects of scientiﬁc practice only make sense on the assumption that scientiﬁc theories are true descriptions of reality. After all, says the realist, if theories are simply convenient calculating devices, then why 294 PHILOSOPHY OF SCIENCE expect different theories to be uniﬁable into one consistent story? Uniﬁcation is clearly desirable if our theories all aim to contribute to the overall truth, but there seems no parallel reason why a bunch of instruments should be uniﬁable into one big ‘instru- ment of everything’. And similarly, the realist will argue, there seems no reason to expect a mere calculating instrument, as opposed to a true description of an underly-	Blackwell
ing reality, to yield a genuine explanation of some past occurrence, or a reliable pre- diction of a future one. However, this form of argument tends to be inconclusive. There are two possible lines of response open to instrumentalists. They can offer an instrumentalist account of the relevant feature of scientiﬁc practice. Alternatively, they can deny that this feature really is part of scientiﬁc practice in the ﬁrst place. As an example of the ﬁrst response, they could argue that the uniﬁcation of science is motivated, not by the pursuit of one underlying truth, but simply by the desirability of having a single all-purpose calculat- ing instrument rather than a rag-bag of different instruments for different problems. The second kind of response would be to deny that uniﬁcation is essential to science to start with. Thus Nancy Cartwright argues that science really is a rag-bag of different instruments. She maintains that scientists faced with a given kind of problem will stan- dardly deploy simplifying techniques and rules of thumb which owe nothing to general theory, but which have shown themselves to deliver the right answer to the kind of problem at hand (Cartwright 1983). Similar responses can be made by instrumentalists to the arguments from explana- tion and prediction. Instrumentalists can either retort that there is no reason why the status of theories as calculating instruments should preclude them from giving rise to predictions and explanations; or they can query whether scientiﬁc theories really do add to our ability to predict and explain to start with. Not all these lines of response are equally convincing. But between them they give instrumentalism plenty of room to counter the initial realist challenge. 1.6 Theory, observation and incommensurability A different line of argument against instrumentalism focuses on the distinction between what is observable and what is not. This distinction is crucial to instrumen- talism, in that instrumentalists argue that claims about observable phenomena are unproblematic, but claims about unobservables are not. However, a number of writers have queried this distinction, arguing that observation reports are not essentially dif- ferent from claims about unobservables, since they too depend on theoretical assump- tions about the underlying structure of reality. Norwood Hanson (1958) has argued, for example, that scientists before and after Copernicus saw different things when they looked at the Sun: whereas pre-Copernicans regarded the Earth as stationary and so saw the Sun revolving round it, post-Copernican scientists saw the Sun as stationary and the Earth as rotating. Similarly, Hanson (1963) argues that the photographic plate which looks like a squiggly mess to a lay observer is seen as displaying a well-deﬁned electron–positron pair by an experienced particle physicist. Examples like these under- mine the distinction between what is observable and what is not, since they show that even judgements made in immediate response to sensory stimulation are inﬂuenced by fallible theories about reality. 295 DAVID PAPINEAU Figure 9.1 The Müller–Lyer. Although all three lines are the same length, the top line, with inward-pointing arrowheads, appears to be shorter, and the bottom line, with outward- pointing arrowheads, appears to be longer, than the ‘neutral’ middle line. Nor is the point restricted to recherché observations of astronomical bodies or sub- atomic particles. Even immediate perceptual judgements about the colour, shape and size of medium-sized physical objects can be shown to depend on theoretical assump- tions implicit in our visual systems. Perhaps the best-known illustration is the Müller–Lyer illusion (see ﬁgure 9.1), which shows how our visual system uses complex assumptions about the normal causes of certain kinds of retinal patterns to draw con- clusions about the geometry of physical ﬁgures. And analogous illusions can be used to demonstrate the presence of other theoretical presuppositions in our visual and other	Blackwell
sensory systems. As I said, in the ﬁrst instance the unclarity of the observable–unobservable distinc- tion counts against instrumentalism rather than realism. After all, it is instrumental- ism, not realism, which needs the distinction, since instrumentalism says that we should be sceptical about unobservable claims, but not observable ones, whereas realism is happy to regard both kinds of claims as belief-worthy, so does not mind if they cannot be sharply distinguished. However, there is another way of responding to doubts about the theory–observa- tion distinction. For note that the arguments against the observable–unobservable distinction do not in fact vindicate the realist belief-worthiness of claims about unob- servables; rather, they attack realism from the bottom up, and undermine the belief- worthiness of claims about observables, by showing that even observational claims depend on fallible theoretical assumptions. Obviously, if there is no observable–unob- servable distinction, then all scientiﬁc claims are in the same boat. But on reﬂection it seems that the boat they all end up in is the instrumentalist boat of sceptical disbelief, not the realist one of general faith in science. A number of inﬂuential recent philosophers of science, most prominently T. S. Kuhn (1962) and Paul Feyerabend (1976), have embraced this conclusion wholeheartedly, and maintained that no judgements made within science, not even observational judge- ments, can claim the authority of established truth. Rather, they argue, once scientists have embraced a theory about the essential nature of their subject matter, such as geo- centrism, or Newtonian dynamics, or the wave theory of light, they will interpret all observational judgements in the light of that theory, and so will never be forced to 296 PHILOSOPHY OF SCIENCE recognize the kind of negative observational evidence that might show them that their theory is mistaken. Kuhn and Feyerabend independently lit on the term incommensu- rable to express the view that there is no common yardstick, in the form of theory- independent observation judgements, which can be used to decide objectively on the worth of scientiﬁc theories. Instead, they argue, decisions on scientiﬁc theories are never due to objective observational evidence, but are always relative to the presuppo- sitions, interests and social milieux of the scientists involved. Kuhn’s and Feyerabend’s blanket relativism has provoked much discussion among philosophers of science, but won few whole-hearted converts. Much of the discussion has focused on the status of observations. Most philosophers of science are prepared to accept that all observational judgements in some sense presuppose some element of theory. But many balk at the conclusion that observations therefore never have any independent authority to decide scientiﬁc questions. After all, they point out, most simple observations, such as that a pointer is adjacent to a mark on a dial, presuppose at most a minimal amount of theory, about rigid bodies, say, and about basic local geometry. Since such minimal theories are themselves rarely at issue in serious scien- tiﬁc debates, this minimal amount of theory-dependence provides no reason why obser- vations of pointer readings should not be used to settle scientiﬁc disputes. If a scientiﬁc theory about the behaviour of gases, say, predicts that a pointer will be at a certain place on a dial, and it is observed not to be, then this decides against the theory about gases. It is not to the point to respond that, in taking the pointer reading at face value, we are making assumptions about rigid bodies and local geometry. For nothing in the debate about gases provides any reason to doubt these assumptions. And this of course is why scientists take such pains to work out what their theories imply about things like pointer readings – since observations of pointer readings do not depend on anything con-	Blackwell
tentious, they will weigh with all sides in the scientiﬁc debate. So, despite the arguments of Kuhn and Feyerabend, nearly all philosophers are real- ists about pointer readings and similar observable phenomena. But this still leaves us with the original disagreement between realism and instrumentalism about less directly observable entities. For, even if claims about pointer readings are uncontroversially belief-worthy, instrumentalists can still argue that theories about viruses, atoms and gravitational waves are nothing more than useful ﬁctions for making calculations. The realist response, as I said, is that the observable facts provide good indirect evi- dence for these theoretical entities, even if we cannot observe them directly. However, there are two strong lines of argument that instrumentalists can use to cast doubt on this suggestion. In the next two sections I shall discuss ‘the underdetermination of theory by evidence’ and ‘the pessimistic meta-induction from past falsity’. 1.7 The underdetermination of theory by evidence The argument from underdetermination asserts that, given any theory about unob- servables that ﬁts the observable facts, there will be other incompatible theories that ﬁt the same facts. And so, the argument concludes, we are never in a position to know that any one of these theories is the truth. Why should we accept that there is always more than one theory that ﬁts any set of observable facts? One popular argument for this conclusion stems from the 297 DAVID PAPINEAU ‘Duhem–Quine thesis’. According to this thesis, any particular scientiﬁc theory can always be defended in the face of contrary observations by adjusting auxiliary hypoth- eses. For example, when the Newtonian theory of gravitation was threatened by obser- vations of anomalous movements by the planet Mercury, it could always be defended by postulating a hitherto unobserved planet, say, or an inhomogeneous mass distribu- tion in the Sun. This general strategy for defending theories against contrary evidence seems to imply that the adherents of competing theories will always be able to main- tain their respective positions in the face of any actual observational data. Another argument for underdetermination starts, not with competing theories, but with some given theory. Suppose that all the predictions of some particular theory are accurate. We can construct a ‘de-Ockhamized’ version of this theory (reversing William of Ockham’s ‘razor’ which prescribes that ‘entities are not to be multiplied beyond necessity’), by postulating some unnecessarily complicated unobservable mechanism which nevertheless yields a new theory with precisely the same observational conse- quences as the original one. Both of these lines of reasoning can be used to argue that more than one theory about unobservables will always ﬁt any given set of observational data. Does this make realism about unobservables untenable? Many philosophers conclude that it does. But this is too quick. For we should recognize that there is nothing in the arguments for alternative underdetermined theories to show that these alternative theories will always be equally well-supported by the data. What the arguments show is that different theories will always be consistent with the data. But they do not rule out the possibility that, among these alternative theories, one is vastly more plausible than the others, and for that reason should be believed to be true. After all, ‘ﬂat earthers’ can make their view consistent with the evidence from geography, astronomy and satellite pho- tographs, by constructing far-fetched stories about conspiracies to hide the truth, the effects of empty space on cameras, and so on. But this does not show that we need take their ﬂat-earthism seriously. Similarly, even though Newtonian gravitational theory can in principle be made consistent with all the contrary evidence, this is no reason not	Blackwell
to believe general relativity. Nor is our ability to ‘cook up’ a de-Ockhamized version of general relativity a reason to stop believing the standard version unencumbered with unnecessary entities. Nevertheless, as I said, many contemporary philosophers of science do move directly from the premise that different theories are consistent with the observational evidence to the conclusion that none of them can be regarded as the truth. This is because many of them address this issue from an essentially Popperian perspective. For if you follow Popper in rejecting induction, then you will not believe that evidence ever provides posi- tive support for any theory, except in the back-handed sense that the evidence can fail to falsify it. Accordingly you will think that all theories that have not been falsiﬁed are on a par, and in particular that any two theories that are both consistent with the evidence are equally well-supported by it. So the arguments for underdetermination do present a problem to Popperians, since Popperians have no obvious basis for discriminating among different theories consis- tent with the data. But, as I pointed out above, these need not worry those of us who diverge from Popper in thinking unfalsiﬁed theories can be better or worse supported by evidence, for we can simply respond to the underdetermination arguments by 298 PHILOSOPHY OF SCIENCE observing that some underdetermined theories are better supported by the evidence than others. Now that we have returned to Popper, it is worth noting that the Duhem–Quine argument also raises a more speciﬁc problem for Popperians. Recall that Popper’s overall philosophy raised the ‘problem of demarcation’, the problem of how to distin- guish science from other kinds of conjecture. Popper’s answer was that science, unlike astrology, or Marxism and psychoanalytic theory, is falsiﬁable. But the Duhem–Quine argument shows that even such eminently scientiﬁc theories as Newtonian physics are not falsiﬁable in any straightforward sense, since they can always save themselves in the face of failed predictions by adjusting auxiliary hypotheses. Not only does this cast doubt on Popper’s dismissal of Marxism and psychoanalysis as unscientiﬁc, but it seems to undermine his whole solution to the demarcation problem. If such paradigmatic scientiﬁc theories as Newtonian physics are not falsiﬁ- able, then it can scarcely be falsiﬁability that distinguishes science from non-science. Still, this is Popper’s problem, not ours (see Harding 1975). If we do not reject induc- tion, then we do not have a problem of demarcation. For we can simply say that what distinguishes successful scientiﬁc theories from non-science is that the observational evidence gives us inductive reason to regard scientiﬁc theories as true. The arguments in the latter part of this section have presupposed that a certain kind of inductive argument is legitimate. The kind of inductive argument relevant to under- determination is not simple ‘enumerative’ induction, from observed As being Bs to ‘All As are Bs’, but rather inferences from any collection of observational data to the most plausible theory about unobservables that is consistent with that data. But these are species of the same genus; indeed, enumerative inductions can themselves be inter- preted as treating ‘All As are Bs’ as the most plausible extrapolation consistent with the observed As being Bs. My attitude to this more general category of inductive inferences remains the same as my attitude to enumerative induction, which I outlined earlier. We do not yet have an explanation of why inductive inferences are legitimate, and to that extent we still face a problem of induction. But it is silly to try to solve that problem by denying that inductive inferences are ever legitimate. And that is why the underdeter- mination of theory by data does not constitute a good argument for instrumentalism. For to assume that we are never entitled to believe a theory, if there are others	Blackwell
consistent with the same data, is simply to assume the illegitimacy of induction. The Underdetermination of Theory by Observational Data (UTD) There are two arguments for the UTD. The ﬁrst is based on the Duhem–Quine thesis, orig- inally formulated by the French philosopher and historian Pierre Duhem (1861–1916) and later revived by the American logician W. V. O. Quine (b. 1908). Duhem (1951) and Quine (1951) point out that a scientiﬁc theory T does not normally imply predictions P on its own, but only in conjunction with auxiliary hypotheses H. T & H ﬁ P So when P is falsiﬁed by observation, this does not refute T, but only the conjunction of T & H. 299 DAVID PAPINEAU not-P ﬁ not-(T & H) So T can be retained, and indeed still explain P, provided we replace H by some alterna- tive, H’, such that T & H’ ﬁ not-P. This yields the Duhem–Quine thesis: any theoretical claim T can consistently be retained in the face of contrary evidence, by making adjustments elsewhere in our system of beliefs. The UTD follows quickly. Imagine two competing theories T1 and T2. Whatever evidence accumulates, versions of T1 and T2, conjoined with greatly revised auxiliary hypotheses if necessary, will both survive, consistent with that evidence, but incompat- ible with each other. The other argument, ﬁrst put forward by physicists like Henri Poincaré (1854–1912) and Ernst Mach (1838–1916) at the turn of the twentieth century, has a different starting-point. Imagine that T1 is the complete truth about physical reality, and that it implies observational facts O. Then we can always construct some ‘de-Ockhamized’ T2 which postulates more complicated unobservable mechanisms but makes just the same observational predictions O. (Glymour 1980: ch. 5.) For example, suppose we start with standard assumptions about the location of bodies in space-time and about the forces acting on them. A de-Ockhamized theory might then postulate that all bodies, including all measuring instruments, are accelerating by 1ft/sec.2 in a given direction, and then add just the extra forces required to explain this. This theory would clearly have exactly the same observational consequences as the original one, even though it contradicted it at the unobservable level. To bring out the difference between the two arguments for UTD, note that the Duhem–Quine argument does not specify exactly which overall theories we will end up with, since it leaves open how T1’s and T2’s auxiliary hypotheses may need to be revised; the de-Ockhamization argument, by contrast, actually speciﬁes T1 and T2 in full detail, including auxiliary hypotheses. In compensation, the Duhem–Quine argument promises us alternative theories whatever observational evidence may turn up in the future; whereas the de-Ockhamization argument assumes that all future observations are as T1 predicts. 1.8 The pessimistic meta-induction I turn now to the other argument against realism. This argument takes as its premise the fact that past scientiﬁc theories have generally turned out to be false, and then moves inductively to the pessimistic conclusion that our current theories are no doubt false too. (This is called a ‘meta-induction’ because its subject matter is not the natural world, but scientiﬁc theories about the natural world.) There are plenty of familiar examples to support this argument. Newton’s theory of space and time, the phlogiston theory of combustion, and the theory that atoms are indivisible were all at one time widely accepted scientiﬁc theories, but have since been recognized to be false. So does it not seem likely, the pessimistic induction concludes, that all our current theories are false, and that we should therefore take an instru- mentalist rather than a realist attitude to them? (See Laudan 1981.) 300 PHILOSOPHY OF SCIENCE This is an important and powerful argument, but it would be too quick to conclude that it discredits realism completely. It is important that the tendency to falsity is much	Blackwell
more common in some areas of science than others. Thus it is relatively normal for theories to be overturned in cosmology, say, or fundamental particle physics, or the study of primate evolution. By contrast, theories of the molecular composition of dif- ferent chemical compounds (such as that water is made of hydrogen and oxygen), or the causes of infectious diseases (chickenpox is due to a herpes virus), or the nature of everyday physical phenomena (heat is molecular motion), are characteristically retained once they are accepted. Nor need we regard this differential success rate of different kinds of theories as some kind of accident. Rather, it is the result of the necessary evidence being more easily available in some areas of science than others. Paleoanthropologists want to know how many hominid species were present on earth 3 million years ago. But their evidence consists of a few pieces of teeth and bone. So it is scarcely surprising that discoveries of new fossil sites will often lead them to change their views. The same point applies on a larger scale in cosmology and particle physics. Scientists in these areas want to answer very general questions about the very small and the very distant. But their evidence derives from the limited range of technological instruments they have devised to probe these realms. So, once more, it is scarcely surprising that their theories should remain at the level of tentative hypotheses. By contrast, in those areas where adequate evidence is available, such as chemistry and medicine, there is no corresponding barrier to science moving beyond tentative hypotheses to ﬁrm conclusions. The moral is that realism is more defensible for some areas of science than others. In some scientiﬁc subjects ﬁrm evidence is available, and entitles us to view certain theories, like the theory that water is composed of H2O molecules, as the literal truth about reality. In other areas the evidence is fragmentary and inconclusive, and then we do better to regard the best-supported theories, such as the theory that quarks and leptons are the ultimate building blocks of matter, as useful instruments which accom- modate the existing data, make interesting predictions, and suggest further lines for research. At ﬁrst sight this might look like a victory for instrumentalism over realism. For did not instrumentalists always accept that we should be realists about observable things, and only urge instrumentalism for uncertain theories about unobservable phenomena? But our current position draws the line in a different place. Instrumen- talism, as originally deﬁned, takes it for granted that everything unobservable is inaccessible, and that all theories about unobservables are therefore uncertain. By contrast, the position we have arrived at places no special weight on the distinction between what is observable and what is not. In particular, it argues that the pessimistic meta-induction fails to show that falsity is the natural fate of all theories about unobservables, but only that there is a line within the category of theories about unobservables, between those theories that can be expected to turn out false and those whose claims to truth are secure. So our current position is not a dogmatic instru- mentalism about all unobservables, but merely the uncontentious view that we should be instrumentalists about that sub-class of theories which are not supported by adequate evidence. 301 DAVID PAPINEAU 1.9 Naturalized epistemology of science In the last decade or so a number of philosophers of science have turned to a natural- ized approach to scientiﬁc knowledge (Kitcher 1992). In place of traditional attempts to establish criteria for scientiﬁc theory-choice by a priori philosophical investigation,	Blackwell
the naturalized approach regards science itself as a subject for a posteriori empirical investigation. Accordingly, naturalized epistemologists look to the history, sociology and psychology of science, rather than to ﬁrst principles, to identify criteria for the acceptability of scientiﬁc theories. One apparent difﬁculty facing this kind of naturalized epistemology of science is that it is unclear how empirical investigation can ever yield anything more than descriptive information about how scientists actually operate. Yet any epistemology of science worth its name ought also to have a normative content – it ought to prescribe how scientists should reason, as well as describe how they do reason. David HUME (chapter 31) ﬁrst pointed out that there is a logical gap between ‘is’ and ‘ought’. A naturalized epistemology based on the empirical study of science seems fated to remain on the wrong side of this gap. However, there is room for naturalized epistemologists to reply to this charge. They can agree that the empirical study of science cannot by itself yield prescriptions about how science ought to be done. But empirical study can still be relevant to such prescriptions. Suppose it is agreed that technological fertility, in the sense of generating technological advances, is a virtue in a scientiﬁc theory. Then the history, sociology and psychology of science might be able to show us that certain kinds of research strategies are effective at developing technologically fertile theories. More generally, given any agreed theoretical end Y, empirical study can show that research strategy X is an effective means to that end. The empirical study of science can thus yield the hypothetical prescription that, if you want Y, then you ought to adopt means X. It is this kind of hypothetical prescription that naturalized philosophers of science seek to establish: they look to the history, sociology and psychology of science to show us that scientists who choose theories on grounds X will in general achieve theories with characteristic Y. Can the naturalized study of science tell us which research strategies are an effec- tive means to theoretical truth? Different naturalized philosophers of science give dif- ferent answers to this question. Many are suspicious of the idea of theoretical truth, and instead prefer to stick to the study of how to achieve more practical ends like tech- nological fertility, simplicity and predictive accuracy. However, there seems no good reason for this restriction. There is nothing obviously incoherent in the idea of look- ing to the empirical study of science to tell us which research strategies have proved a good way of developing true theories. Indeed, the discussion of the ‘pessimistic meta- induction’ in the previous section amounted to the sketch of just such an investigation, in that it appealed to the history of science to decide whether or not the standard pro- cedures of scientiﬁc theory-choice succeed in identifying true theories. It is not difﬁcult to imagine more detailed and speciﬁc studies of this kind of issue. Let me now return brieﬂy to the issue with which I began, namely, the problem of induction. It is possible that the naturalized study of how to get at the scientiﬁc truth will enable us to make headway with this problem. For an empirical investigation into 302 PHILOSOPHY OF SCIENCE science might be able to show us that a certain kind of inductive inference is in general a reliable guide to scientiﬁc truth. And this would then provide a kind of vindication of that inductive method (see Papineau 1993: ch. 5). It is true that this kind of defence of induction will inevitably involve an element of circularity. For when we infer that certain kinds of induction are in general a reliable guide to truth, on the basis of evidence from the history of science, this will itself be an inductive inference. It is a matter of some delicacy, however, whether this circularity is vicious.	Blackwell
Defenders of this naturalized defence of induction will point out that, from their point of view, a legitimate criterion of theory-choice need not be an a priori guide to truth, but only an empirically certiﬁable one. Given this, the original argument against induction, that it is not logically valid, will not worry naturalized philosophers of science. Induction may not provide any a priori guarantee for its conclusions; but from the naturalized point of view, this does not show that induction is in any way illegiti- mate, since it leaves it open that induction may be an empirically reliable guide to the truth. And if there is nothing to show that induction is illegitimate, naturalized philoso- phers of science can then argue, why should we not use it to investigate the worth of inductive inferences? Maybe this is less satisfying a defence of induction than we might originally have hoped for. But perhaps it is defence enough. 2 The Metaphysics of Science 2.1 Causation Many issues in the metaphysics of science hinge on the notion of causation. This notion is as important in science as it is in everyday thinking, and much scientiﬁc theorizing is concerned speciﬁcally to identify the causes of various phenomena. However, there is little philosophical agreement on what it means to say that one event is the cause of another. Modern discussion of causation starts with David Hume, who argued that causation is simply a matter of CONSTANT CONJUNCTION (p. 720). According to Hume (1978), one event causes another if and only if events of the type to which the ﬁrst event belongs regularly occur in conjunction with events of the type to which the second event belongs. This formulation, however, leaves a number of questions open. Firstly, there is the problem of distinguishing genuine causal laws from accidental regularities. Not all regularities are sufﬁciently lawlike to underpin causal relationships. Being a screw in my desk could well be constantly conjoined with being made of copper, without its being true that these screws are made of copper because they are in my desk. Secondly, the idea of constant conjunction does not give a direction to causation. Causes need to be distinguished from effects. But knowing that A-type events are con- stantly conjoined with B-type events does not tell us which of A and B is the cause and which the effect, since constant conjunction is itself a symmetric relation. Thirdly, there is a problem about probabilistic causation. When we say that causes and effects are constantly conjoined, do we mean that the effects are always found with the causes, or is it enough that the causes make the effects probable? 303 DAVID PAPINEAU Many philosophers of science during the past century have preferred to talk about explanation rather than causation. According to the covering-law model of explanation, something is explained if it can be deduced from premises which include one or more laws. As applied to the explanation of particular events, this implies that one particu- lar event can be explained if it is linked by a law to some other particular event. However, while they are often treated as separate theories, the covering-law account of explanation is at bottom little more than a variant of Hume’s constant conjunction account of causation. This afﬁnity shows up in the fact that the covering-law account faces essentially the same difﬁculties as Hume: (1) in appealing to deductions from ‘laws’, it needs to explain the difference between genuine laws and accidentally true regularities; (2) it omits the requisite directionality, in that it does not tell us why we should not ‘explain’ causes by effects, as well as effects by causes; after all, it is as easy to deduce the height of a ﬂagpole from the length of its shadow and the laws of optics, as to deduce the length of the shadow from the height of the pole and the same laws;	Blackwell
(3) are the laws invoked in explanation required to be exceptionless and deterministic, or is it acceptable, say, to appeal to the merely probabilistic fact that smoking makes cancer more likely, in explaining why some particular person developed cancer? In what follows I shall discuss these three problems in order (treating them as prob- lems that arise equally both for the analysis of causation and the analysis of explana- tion). After that I shall consider some further issues in the metaphysics of science. The Covering-Law Model of Explanation According to this model (originally proposed by Hempel and Oppenheim (1948) and further elaborated in Hempel (1965)) one statement (the explanandum) is explained by other statements (the explanans) if and only if the explanans contains one or more laws, and the explanandum can be deduced from the explanans. In the simplest case, where the explanandum is some particular statement to the effect that some individual a has property E, we might therefore have: a has C For all x, if x has C, then x has E a has E For example, we might deduce that a piece of litmus paper turned red, from the law that all litmus paper placed in acid turns red, together with the prior condition that this piece of litmus paper was in fact placed in acid. The model can accommodate more compli- cated explanations of particular events, and can also allow explanations of laws themselves, as when we deduce Kepler’s law that all planets move in ellipses, say, from Newton’s law of universal gravitation and his laws of motion. As applied to the explanation of particular events, the covering-law model implies a symmetry between explanation and prediction. For the information that, according to the model, sufﬁces for the explanation of some known event should also enable us to predict that event if we did not yet know of it. Many critics have fastened on this impli- cation of the model, however, and pointed out that we can often predict when we do not have enough information to explain (as when we predict the height of the ﬂagpole from 304 PHILOSOPHY OF SCIENCE its shadow) and can often explain when we could not have predicted (as when we explain X’s cancer on the basis of X’s smoking). These examples suggest that genuine explanations of particular events need to cite genuine causes, and that the reason the covering-law model runs into counter-examples is that it adds nothing to the inadequate constant conjunction analysis of causation, except that it substitutes the term ‘law’ for ‘constant conjunction’. To get a satisfactory account of explanation we need, ﬁrstly, to recognize that explanations of particular events must mention causes, and, secondly, to improve on the constant conjunction analysis of causation. There is a variant of the covering-law model which allows non-deterministic expla- nation as well as deterministic ones. This is termed the ‘inductive–statistical (I–S)’ model, by contrast with the original ‘deductive–nomological (D–N)’ model. An example would be: a drinks 10 units of alcohol per diem For p per cent of xs, if x drinks 10 units of alcohol per diem, x has a damaged liver a has a damaged liver Here the explanandum cannot be deduced from the explanans, but only follows with an inductive probability of p; and the inference appeals to a statistical regularity, rather than an exceptionless nomological generalization. In Hempel’s original version of this model, it was required that the probability of the explanandum be high. A better requirement, however, as explained in the section on probabilistic causation below, is that the par- ticular facts in the explanans need only make the probability of the explanandum higher than it would otherwise have been. 2.2 Laws and accidents There are two general strategies for distinguishing laws from accidentally true gener- alizations. The ﬁrst stands by Hume’s idea that causal connections are mere constant conjunctions, and then seeks to explain why some constant conjunctions are better	Blackwell
than others. That is, this ﬁrst strategy accepts the principle that causation involves nothing more than certain events always happening together with certain others, and then seeks to explain why some such patterns – the ‘laws’ – matter more than others – the ‘accidents’. The second strategy, by contrast, rejects the Humean presupposition that causation involves nothing more than happenstantial co-occurrence, and instead postulates a relationship of ‘necessitation’, a kind of ‘cement’, which links events that are connected by law, but not those events (like being a screw in my desk and being made of copper) that are only accidentally conjoined. There are a number of versions of the ﬁrst Humean strategy. The most successful, originally proposed by F. R. Ramsey (1903–30), and later revived by David Lewis (1973), holds that laws are those true generalizations that can be ﬁtted into an ideal system of knowledge. The thought here is that the laws are those patterns that are somehow explicable in terms of basic science, either as fundamental principles them- selves, or as consequences of those principles, while accidents, although true, have no such explanation. Thus, ‘All water at standard pressure boils at 100°C’ is a consequence 305 DAVID PAPINEAU of the laws governing molecular bonding; but the fact that ‘All the screws in my desk are copper’ is not part of the deductive structure of any satisfactory science. Ramsey neatly encapsulated this idea by saying that laws are ‘consequences of those proposi- tions which we should take as axioms if we knew everything and organized it as simply as possible in a deductive system’ (Ramsey 1978: 130). Advocates of the alternative non-Humean strategy object that the difference between laws and accidents is not a linguistic matter of deductive systematization, but rather a metaphysical contrast between the kind of links they report. They argue that there is a link in nature between being at 100°C and boiling, but not between being in my desk and being made of copper, and that this is nothing to do with how the descrip- tion of this link may ﬁt into theories. According to D. M. Armstrong (1983), the most prominent defender of this view, the real difference between laws and accidents is simply that laws report relationships of natural necessitation, while accidents only report that two types of events happen to occur together. Armstrong’s view may seem intuitively plausible, but it is arguable that the notion of necessitation simply restates the problem, rather than solving it. Armstrong says that necessitation involves something more than constant conjunction: if two events are related by necessitation, then it follows that they are constantly conjoined; but two events can be constantly conjoined without being related by necessitation, as when the constant conjunction is just a matter of accident. So necessitation is a stronger rela- tionship than constant conjunction. However, Armstrong and other defenders of this view say very little about what this extra strength amounts to, except that it distin- guishes laws from accidents. Armstrong’s critics argue that a satisfactory account of laws ought to cast more light than this on the nature of laws. 2.3 The direction of causation Hume said that the earlier of two causally related events is always the cause, and the later the effect. However, there are a number of objections to using the earlier–later ‘arrow of time’ to analyse the directional ‘arrow of causation’. For a start, it seems in principle possible that some causes and effects could be simultaneous. More seriously, the idea that time is directed from ‘earlier’ to ‘later’ itself stands in need of philosophi- cal explanation – and one of the most popular explanations is that the idea of ‘move- ment’ from earlier to later depends on the fact that cause–effect pairs always have a given orientation in time. However, if we adopt such a ‘causal theory of the arrow of	Blackwell
time’, and explain ‘earlier’ as the direction in which causes lie, and ‘later’ as the direction of effects, then we will clearly need to ﬁnd some account of the direction of causation which does not itself assume the direction of time. A number of such accounts have been proposed. David Lewis (1979) has argued that the asymmetry of causation derives from an ‘asymmetry of overdetermination’. The overdetermination of present events by past events – consider a person who dies after simultaneously being shot and struck by lightning – is a very rare occurrence. By contrast, the multiple ‘overdetermination’ of present events by future events is absolutely normal. This is because the future, unlike the past, will always contain mul- tiple traces of any present event. To use Lewis’s example, when the president presses the red button in the White House, the future effects do not only include the dispatch 306 PHILOSOPHY OF SCIENCE of nuclear missiles, but also his ﬁngerprint on the button, his trembling, the further depletion of his gin bottle, the recording of the button’s click on tape, the emission of light waves bearing the image of his action through the window, the warming of the wire from the passage of the signal current, and so on, and on, and on. Lewis relates this asymmetry of overdetermination to the asymmetry of causation as follows. If we suppose the cause of a given effect to have been absent, then this implies the effect would have been absent too, since (apart from freaks like the light- ning–shooting case) there will not be any other causes left to ‘ﬁx’ the effect. By con- trast, if we suppose a given effect of some cause to have been absent, this does not imply the cause would have been absent, for there are still all the other traces left to ‘ﬁx’ the cause. Lewis argues that these counterfactual considerations sufﬁce to show why causes are different from effects. Other philosophers appeal to a probabilistic variant of Lewis’s asymmetry. Follow- ing Reichenbach (1956), they note that the different causes of any given type of effect are normally probabilistically independent of each other; by contrast, the different effects of any given type of cause are normally probabilistically correlated. For example, both obesity and high excitement can cause heart attacks, but this does not imply that fat people are more likely to get excited than thin ones; on the other hand, the fact that both lung cancer and nicotine-stained ﬁngers can result from smoking does imply that lung cancer is more likely among people with nicotine-stained ﬁngers. So this account distinguishes effects from causes by the fact that the former, but not the latter, are probabilistically dependent on each other. 2.4 Probabilistic causation The just-mentioned probabilistic account of the direction of causation is normally for- mulated as part of a more general theory of probabilistic causation. Until relatively recently philosophers assumed that the world fundamentally conforms to determinis- tic laws, and that probabilistic dependencies between types of events, such as that between smoking and lung cancer, merely reﬂected our ignorance of the full causes. The rise of quantum mechanics, however, has persuaded most philosophers that deter- minism is false, and that some events, like the decay of a radium atom, happen purely as a matter of chance. A particular radium atom may decay, but on another occasion an identical atom in identical circumstances might well not decay. Accordingly, a number of philosophers of science have put forward models of cau- sation which require only that causes probabilify, rather than determine, their effects. The earliest such model was the ‘inductive–statistical’ version of the covering-law model of explanation (Hempel 1965). Unlike deterministic ‘deductive–nomological’ explanations, such inductive–statistical explanations required only that prior condi-	Blackwell
tions and laws imply a high probability for the event to be explained, not that this event will certainly happen. However, even this seems too strong a requirement for probabilistic causation. After all, smoking unequivocally causes lung cancer, but even heavy smokers do not have a high probability of lung cancer, in the sense of a proba- bility close to 1. Rather, their smoking increases their probability of lung cancer, not to a high ﬁgure, but merely from a low to a less low ﬁgure, but still well below 50 per cent. So more recent models of probabilistic causation simply require that causes 307 DAVID PAPINEAU should increase the probability of their effects, not that they should give them a high probability (Salmon 1971). This kind of model needs to guard against the possibility that the probabilistic asso- ciation between putative cause and putative effect may be spurious, like the probabilis- tic association between barometers falling and subsequent rain. Such associations are not due to a causal connection between barometer movements and rain, but rather to both of these being joint effects of a common cause, namely, in our example, falls in atmospheric pressure. The obvious response to this difﬁculty is to say that we have a cause–effect relationship between A and B if and only if A increases the probability of B, and this association is not due to some common cause C. However, this is obviously incomplete as an analysis of causation, since it uses the notion of (common) cause in explaining causation. It would solve this problem if we could analyse the notion of common cause in prob- abilistic terms. It seems to be a mark of common causes that they probabilistically ‘screen off’ the associations between their joint effects, in the sense that, if we consider cases where the common cause is present and where it is absent separately, then the probabilistic association between the joint effects will disappear. For example, if it is given that the atmospheric pressure has fallen, then a falling barometer does not make it any more likely that it will rain; and similarly, if the atmospheric pressure has not fallen, a faulty falling reading on a barometer is no probable indicator of impending rain. (Numerically, if C is a common cause, and A and B its joint effects, we will ﬁnd that A and B are associated – Prob(B/A) < Prob(B) – but that C and its absence render A irrelevant to B – Prob(B/A & C) = Prob(B/C) and Prob(B/A & not-C) = Prob(B/not- C).) It remains a matter of some debate, however, whether this characteristic probabilistic structure of common causes is enough to allow a complete explanation of causation in probabilistic terms, or whether further non-probabilistic considerations need to be introduced. 2.5 Probability Philosophical interest in probabilistic causation has led to a resurgence of interest in the philosophy of probability itself. Probability raises philosophical puzzles in its own right, quite apart from its connection with causation. What exactly is the ‘probability’ of a given event? The only part of the answer that is uncontroversial is that probabil- ities are quantities that satisfy the axioms of the probability calculus I speciﬁed earlier when discussing Bayesianism. But this leaves plenty of room for alternative philoso- phical views, for there are a number of different ways of interpreting these axioms. One interpretation is the subjective theory of probability, which equates probabil- ities with subjective degrees of belief. This is the interpretation assumed by Bayesian conﬁrmation theory. Most philosophers are happy to agree that subjective degrees of belief exist, and that the Dutch Book Argument (see the above box on Bayesianism) shows why they ought to conform to the axioms of probability. But many, if not all, philosophers argue that we need a theory of objective probability in addition to this	Blackwell
subjective account. One possible objective interpretation is the frequency theory, originally put forward by Richard von Mises (1957). According to this theory, the probability of a given kind 308 PHILOSOPHY OF SCIENCE of result is the number of times this result occurs, divided by the total number of occa- sions on which it might have occurred. So, for example, the probability of heads on a coin toss is the proportion of heads in some wider class of coin tosses. This theory, however, faces a number of difﬁculties. For a start, it has problems in dealing with ‘single-case probabilities’. Consider a particular coin toss. We can consider it as a member of the class of all coin tosses, or of all tosses of coins with that particular shape, or of all tosses made in just that way, or so on. However, these different ‘reference classes’ may well display different frequencies of heads. Yet intuitively it seems that there ought to be a unique value for the probability of heads on a particular toss of a particular coin. Perhaps this difﬁculty can be dealt with by specifying that the single-case probability should equal the relative frequency in the reference class of all tosses that are similar in all relevant respects to the particular toss in question. But there remain difﬁculties about which respects should count as ‘relevant’ in this sense. In addition, there is the problem that many of these more speciﬁc reference classes will only be ﬁnite in extent. Coins with a certain distinctive shape may only be tossed in some given way ten times in the whole history of the universe. Yet the probability of heads on these tosses is unlikely to be equal to the relative frequency in the ten tosses, for luck may well yield a disproportionately high, or low, number of heads in ten tosses. Because of this, frequency theorists standardly appeal, not to actual reference classes, but to hypothetical inﬁnite sequences, and equate the probability with the limit that the relative frequency would tend to if the relevant kind of trial were repeated an inﬁnite number of times. Critics of the frequency theory object that this reliance on hypotheti- cal inﬁnite reference sequences makes probabilities inadmissibly abstract. Because of these difﬁculties, many contemporary philosophers of probability have adopted the ‘propensity’ theory of probability in place of the frequency theory. The ear- liest version of this theory, proposed by Popper (1959b), simply modiﬁed the frequency theory by specifying that only those relative frequencies generated by repeated trials on a given ‘experimental set-up’ should count as genuine probabilities. This arguably deals with the problem of single-case probabilities, but it still leaves us with hypothetical reference classes. To avoid this, later versions of the propensity theory do not deﬁne probabilities in terms of frequencies at all, but simply take probabilities to be primitive propensities of particular situations to produce given results. This kind of propensity theory does not seek to deﬁne objective probabilities in terms of frequencies, but in effect simply takes single-case probabilities as primitive (see Mellor 1971). But it can still recognize a connection between probabilities and frequencies. For, as long as propensities are assumed to obey the axioms of the probability calculus (though this assumption itself merits some debate), it will follow that, in a sufﬁciently long sequence of independent trials in each of which the propensity to produce B is p, the overall propensity for the observed frequency of B to differ by more than a given amount from p can be made arbitrarily small, in accord with the Law of Large Numbers. (Note how the italicized second use of ‘propensity’ in this claim prevents it serving as a deﬁnition of propensities in terms of frequencies.) Both the frequency theory and the propensity theory have their strengths. The fre-	Blackwell
quency theory has the virtue of offering an explicit deﬁnition of probability, where the propensity theory takes probabilities as primitive. On the other hand, the propensity 309 DAVID PAPINEAU theory has no need to assume hypothetical reference sequences, whereas these are essential to the frequency theory. It may seem that the frequency theory, because it offers an explicit deﬁnition, is better able than the propensity theory to explain how we ﬁnd out about probabilities. But this is an illusion. The trouble is that the frequency theory’s explicit deﬁnition is in terms of frequencies in INFINITE SEQUENCES (p. 355). But our evidence is always in the form of fre- quencies in ﬁnite samples. So the problem of explaining how we can move from fre- quencies in ﬁnite samples to knowledge of probabilities is as much a problem for the frequency theory as for the propensity theory. (There are various suggestions about how to solve this ‘problem of statistical inference’, none of them universally agreed. My present point is merely that this problem of statistical inference arises in just the same way for both frequency and propensity theorists.) In the face of continued debate about the interpretation of objective probability, some philosophers have turned to physics, and in particular to the notion of probabil- ity used in quantum mechanics, to resolve the issue. Unfortunately, quantum mechan- ics is no less philosophically controversial than the notion of probability. There are different philosophical interpretations of the formal theory of quantum mechanics, each of which involves different understandings of probability. Because of this it seems likely that philosophical disputes about probability will continue until there is an agreed interpretation of quantum mechanics. The Interpretation of Quantum Mechanics Modern quantum mechanics says that the state of any given system of microscopic par- ticles is fully characterized by its ‘wave function’. However, instead of specifying the exact positions and velocities of the particles, as is done in classical mechanics, this ‘wave func- tion’ only speciﬁes probabilities of the particles displaying certain values of position, and velocity, if appropriate measurements are made. Schrödinger’s equation then speciﬁes how this wave function evolves smoothly and deterministically over time, analogously to the way that Newton’s laws of motion specify how the positions and velocities of macro- scopic objects evolve over time – except that Schrödinger’s equation again only describes changes in probabilities, not exact values. On the orthodox interpretation of quantum mechanics, quantum probabilities change into actualities only when ‘measurements’ are made. If you measure the posi- tion of a particle, say, then its position assumes a deﬁnite value, even though nothing before the measurement determined exactly what this value would be. There is something puzzling about this, however, since any overall system of mea- sured particle and measuring instrument is itself just another system of microscopic par- ticles, which might therefore be expected to evolve smoothly according to Schrödinger’s equation, rather than to jump suddenly to some deﬁnite value for position. To account for this, the orthodox interpretation says that in addition to the normal Schrödinger evo- lution, there is a special kind of change which occurs in ‘measurements’, when the wave function suddenly ‘collapses’ to yield a deﬁnite value for the measured quantity. The ‘measurement problem’ is the problem of explaining exactly when, and why, these collapses occur. The story of ‘Schrödinger’s cat’ makes the difﬁculty graphic. Suppose that some unfortunate cat is sitting next to a poison dispenser which is wired 310 PHILOSOPHY OF SCIENCE up to emit cyanide gas if an electron emitted from some source turns up on the right half of some position-registering plate, but not if the electron turns up on the left half. The	Blackwell
basic quantum mechanical description of this situation says that it is both possible that the electron will turn up on the right half of the plate and that it will turn up on the left half, and therefore both possible that the poison is emitted and that it is not, and there- fore both possible that the cat is alive and that it is dead. One of these possibilities only becomes actual when the wave function of the whole system collapses. But when does that happen? When the electron is emitted? When it reaches the plate? When the cat dies or not? Or only when a human being looks at the cat to see how it is faring? There seems no principled way to decide between these answers. Because of this, many philosophers reject the orthodox view that physical systems are completely char- acterized by their wave functions, and conjecture that, in addition to the variables quantum mechanics recognizes, there are various ‘hidden variables’ which always specify exact positions and velocities for all physical particles. It is difﬁcult, however, for such hidden variable theories to reproduce the surprising phenomena predicted by quantum mechanics, without postulating mysterious mechanisms that seem inconsis- tent with other parts of physics. A more radical response to the measurement problem is to deny that the wave func- tion ever does collapse, and somehow to make sense of the idea that reality contains both a live cat and a dead cat. This ‘many-worlds’ interpretation of quantum mechanics ﬂies in the face of common sense, but its theoretical attractions are leading an increasing number of philosophers to take it seriously. 2.6 Teleology We normally explain some particular fact by citing its cause: for example, we explain why some water freezes by noting that its temperature fell below 0°C. There are cases, however, where we seem to explain items by citing their effects instead. In particular, this kind of explanation is common in biology. We often explain some biological trait by showing how it is useful to the organism in question: for instance, the explanation of the polar bear’s white fur is that it camouﬂages it; the explanation of human sweating is that it lowers body temperature, and so on. Similar explanations are also sometimes offered in anthropology and sociology. Until fairly recently most philosophers of science took such functional or teleologi- cal explanations at face value, as an alternative to causal explanation, in which items are explained, not by their causal antecedents, but by showing how they contribute to the well-being of some larger system. Carl Hempel’s covering-law model of explanation embodied an inﬂuential version of this attitude. According to Hempel, causal explana- tions and functional explanations are simply two different ways of exemplifying the covering-law model: the only difference is that in causal explanations the explaining fact (lower temperature) temporally precedes the explained fact (freezing), whereas in func- tional explanations it is explained fact (white fur) that comes temporally before the con- sequence (camouﬂage) which explains it. Most contemporary philosophers of science, however, take a different view, and argue that all explanations of particular facts are really causal, and that functional explanations, despite appearances, are really a subspecies of causal explanations. On 311 DAVID PAPINEAU this view, the reference to future facts in functional explanations is merely ap- parent, and such explanations really refer to past causes. In the biological case, these past causes will be the evolutionary histories that led to the natural selection of the biological trait in question. Thus the functional explanation of the polar bear’s colour should be understood as referring us to the fact that their past camouﬂaging led to the natural selection of their whiteness, and not to the fact that they may	Blackwell
be camouﬂaged in the future. Similarly, any acceptable functional explanations in anthropology or sociology should be understood as referring us back in time to the conscious intentions or unconscious selection processes which caused the facts to be explained (see Wright 1973; Neander 1991). (There remains the terminological matter of whether functional explanations understood in this way ought still to be called ‘tele- ological’. Traditional usage reserves the term ‘teleology’ for distinctively non-causal explanations in terms of future results. But most contemporary philosophers are happy to describe disguised causal explanations that make implicit reference to selection mechanisms as ‘teleological’.) 2.7 Theoretical reduction Another philosophical question about subject matters like biology is whether they can be reduced to lower-level (in the sense of ontologically more basic) sciences like chem- istry and physics. Obviously, this is an issue that arises not just for biology, but also for such other ‘special’ natural sciences as geology and meteorology, and also for such human sciences as psychology, sociology and anthropology. One science is said to ‘reduce’ to another if its categories can be deﬁned in terms of the categories of the latter, and its laws explained by the laws of the latter. Reduction- ists argue that all sciences form a hierarchy in which the higher can always be reduced to the lower. Thus, for example, biology might be reduced to physiology, physiology to chemistry, and eventually chemistry to physics. Reductionism can be viewed either historically or metaphysically. The historical question is whether science characteristically progresses by later theories reducing earlier ones. The metaphysical question is whether the different areas of science describe different realities, or just the one physical reality described at different levels of detail. Though often run together, these are different questions. Taken as a general thesis, historical reductionism is false. Recall the earlier discus- sion of the ‘pessimistic meta-induction from past falsity’. This involved the claim that new theories characteristically show their predecessors to be false. To the extent that this claim is true, historical reductionism is false: for a new theory can scarcely explain why an earlier theory was true, if it shows it is false. In the earlier discussion I argued that there are some areas of science, like molecular biology and medical science, to which the pessimistic meta-induction does not apply. If this is right, then we can expect that in these areas new theories will indeed normally reduce old ones. But I did not dispute that there are other areas of science, like cosmology and fundamental particle physics, in which the normal fate of old theories is to be thrown out. It follows that we must reject historical reductionism, understood as the thesis that all science proceeds by new theories reduc- ing old ones. 312 PHILOSOPHY OF SCIENCE This does not mean, however, that metaphysical reductionism is false. Even if science proceeds towards the overall truth by ﬁts and starts, there may be general reasons for expecting that this overall truth, when eventually reached, will reduce to physical truth. One possible such argument stems from the causal interaction between the phenom- ena discussed in the special sciences and physical phenomena. Biological, geological and meteorological events all unquestionably have physical effects. It is difﬁcult to see how they could do this unless they are made of physical components. It is doubtful, however, whether this sufﬁces to establish full-scale reductionism, as opposed to the weaker thesis (sometimes called ‘token-identity’) according to which each particular higher-level event is identical with some particular physical event. Thus, for example, it might be true that one animal’s aggressive behaviour can be equated	Blackwell
with a given sequence of physical movements, and another animal’s aggressive behav- iour can be equated with another sequence of physical movements, without there being any uniform way of deﬁning ‘aggressive behaviour’, for all animals, in terms of physi- cal movements. The case-by-case token-identity will explain how each instance of aggressive behaviour can have physical effects, like causing intruding animals to move away. But without any uniform deﬁnition of ‘aggressive behaviour’ in terms of physi- cal movements there is no question of reducing ethology (the science of behaviour) to physics, and so no question of explaining ethological laws by physical laws. Instead, the laws of ethology and other special sciences will be sui generis, identifying patterns whose instances vary in their physical make-up, and which therefore cannot possibly be explained in terms of physical laws alone (see Fodor 1974). Acknowledgements I would like to thank Stathis Psillos for helping me with this chapter. Further Reading For an introduction to the problem of induction, and Popper’s solution, see Popper (1959a), espe- cially chapter 1. The problem and Popper’s solution are further discussed in O’Hear (1989). There are two excellent introductions to Bayesian philosophy of science: Horwich (1982) and Howson and Urbach (1989). The best modern defence of instrumentalism is Van Fraassen (1980). Churchland and Hooker (1985) offers a good collection of essays on the realism–instrumentalism debate. Kitcher (1993) contains a strong defence of realism. The classic works on the theory-dependence of observation and the incommensurability of theories are Hanson (1958), Kuhn (1962) and Feyerabend (1976). A good collection of essays on these issues is Hacking (1981). The best sources for the underdetermination of theories by evidence and the pessimistic meta-induction are respectively Quine (1951) and Laudan (1981). For a survey of recent work on naturalized epistemology, see Kitcher (1992). Most modern discussions of explanation begin with the title essay in Hempel (1965). Expla- nation and its relation to causation are further explored by the essays in Ruben (1993). Arm- strong (1983) provides an excellent account of the general problem of distinguishing laws from accidents, as well as his own solution. Chapter 7 of O’Hear (1989) contains a good introduction to both probability and probabilistic causation. The best contemporary non-specialist discussion of the problems of quantum mechanics is to be found in chapters 11–13 of Lockwood (1989). 313 DAVID PAPINEAU The view that events discussed in the special sciences are token-identical but not reducible to physical events is defended in Fodor (1974). References Armstrong, D. 1983: What is a Law of Nature? Cambridge: Cambridge University Press. Ayer, A. J. 1956: The Problem of Knowledge. London: Macmillan. Boyd, R. 1980: Scientiﬁc Realism and Naturalistic Epistemology. In P. Asquith and R. Giere (eds) PSA 1980 vol. 2, 613–62. East Lansing, MI: Philosophy of Science Association. Cartwright, N. 1983: How the Laws of Physics Lie. Oxford: Oxford University Press. Churchland, P. and Hooker, C. (eds) 1985: Images of Science. Chicago: University of Chicago Press. Duhem, P. 1951 [1906]: The Aim and Structure of Physical Theory (translated by P. Wiener). Princeton, NJ: Princeton University Press. Feyerabend, P. 1976: Against Method. London: New Left Books. Fodor, J. 1974: Special Sciences. Synthèse, 28, 97–115. Friedman, M. 1984: Foundations of Spacetime Theories. Princeton, NJ: Princeton University Press. Glymour, C. 1980: Theory and Evidence. Princeton, NJ: Princeton University Press. Hacking, I. 1967: Slightly More Realistic Personal Probability. Philosophy of Science, 34, 311–25. —— (ed.) 1981: Scientiﬁc Revolutions. Oxford: Oxford University Press. Hanson, N. R. 1958: Patterns of Discovery. Cambridge: Cambridge University Press. —— 1963: The Concept of the Positron. Cambridge: Cambridge University Press. Harding, S. (ed.) 1975: Can Theories be Refuted? Dordrecht: Reidel.	Blackwell
Hempel, C. 1965: Aspects of Scientiﬁc Explanation. New York: Free Press. Hempel, C. and Oppenheim, P. 1948: Studies in the Logic of Explanation. Philosophy of Science, 15, 135–75. Horwich, P. 1982: Probability and Evidence. Cambridge: Cambridge University Press. Howson, C. and Urbach, P. 1989: Scientiﬁc Reasoning. La Salle, IN: Open Court. Hume, D. 1978 [1739]: A Treatise of Human Nature (edited by P. H. Nidditch). Oxford: Clarendon Press. Kitcher, P. 1992: The Naturalists Return. Philosophical Review, 101, 53–114. —— 1993: The Advancement of Science. New York: Oxford University Press. Kuhn, T. S. 1962: The Structure of Scientiﬁc Revolutions. Chicago: Chicago University Press. Laudan, L. 1981: A Confutation of Convergent Realism. Philosophy of Science, 48. Lewis, D. 1973: Counterfactuals. Oxford: Blackwell. —— 1979: Counterfactual Dependence and Time’s Arrow. Noûs, 13, 455–76. Lockwood, M. 1989: Mind, Brain and the Quantum. Oxford: Blackwell. Mellor, D. 1971: The Matter of Chance. Cambridge: Cambridge University Press. Mises, R. von 1957: Probability, Statistics and Truth. London: Allen and Unwin. Neander, K. 1991: The Teleological Notion of Function. Australasian Journal of Philosophy, 69, 454–68. O’Hear, A. 1989: An Introduction to the Philosophy of Science. Oxford: Clarendon Press. Papineau, D. 1993: Philosophical Naturalism. Oxford: Blackwell. Popper, K. 1959a: The Logic of Scientiﬁc Discovery. London: Hutchinson. —— 1959b: The Propensity Interpretation of Probability. British Journal for the Philosophy of Science, 10, 25–42. —— 1963: Conjectures and Refutations. London: Routledge and Kegan Paul. —— 1972: Objective Knowledge. Oxford: Clarendon Press. Quine, W. V. O. 1951: Two Dogmas of Empiricism. In From a Logical Point of View. New York: Harper. 314 PHILOSOPHY OF SCIENCE Ramsey, F. 1978 [1929]: General Propositions and Causality. In D. H. Mellor (ed.) Foundations. London: Routledge and Kegan Paul. Reichenbach, H. 1956: The Direction of Time. Berkeley: University of California Press. Ruben, D.-H. (ed.) 1993: Explanation. Oxford: Oxford University Press. Salmon, W. 1971: Statistical Explanation and Statistical Relevance. Pittsburgh, PA: University of Pittsburgh Press. Smart, J. 1963: Philosophy and Scientiﬁc Realism. London: Routledge and Kegan Paul. Teller, P. 1973: Conditionalization and Observation. Synthèse, 28, 218–58. Van Fraassen, B. 1980: The Scientiﬁc Image. Oxford: Clarendon Press. Worrall, J. 1989: Why Both Popper and Watkins Fail to Solve the Problem of Induction. In F. D’Agostino and I. Jarvie (eds) Freedom and Rationality. Dordrecht: Kluwer. Wright, L. 1973: Functions. Philosophical Review, 82, 139–68. Discussion Questions 1 Is it any less rational to accept induction than it is to accept deduction? 2 Is it always a mistake to save a theory when it has been falsiﬁed? 3 How can we show that it is more rational to believe some hypotheses than to believe others? 4 How do scientiﬁc theories move beyond the stage of conjecture? 5 Is belief a matter of degree? 6 Does Bayes’s theorem help us to deal rationally with evidence? 7 Do scientiﬁc claims about unobservable entities differ in status from scientiﬁc claims about observable entities? 8 Is instrumentalism mistaken if it cannot account for some features of scientiﬁc practice? How can we determine what features are really a part of scientiﬁc practice? 9 Are general theories or piecemeal procedures more important to our basic characterization of science? 10 Does all observation depend on theoretical assumptions? What are the implica- tions of your answer for an account of unobservables? 11 ‘Given any theory about unobservables which ﬁts the observed facts, there will always be other incompatible theories which ﬁt the same facts.’ Does this make realism about unobservables untenable? 12 Can we be realists in some areas of science and instrumentalists in others? 13 Should we look to the history, sociology, and psychology of science, rather than	Blackwell
to ﬁrst principles, to identify criteria for the acceptability of scientiﬁc theories? 14 Can a naturalized study of science tell us which research strategies are an effective means to theoretical truth? 15 Is there any good reason for a philosopher to prefer to talk about explanation rather than causation? 16 How can we distinguish laws from accidentally true generalizations? 17 Must we posit an ideal system of knowledge in order to understand the notion of ‘law’? What if there cannot be such a system? 18 Do scientiﬁc laws involve necessity? In what sense? 19 How can we explain the direction of causation? 20 Can we accept a model of causation according to which causes probabilify, rather than determine, their effects? 315 DAVID PAPINEAU 21 What problem poses greater difﬁculties for frequency theories of probability: ‘single-case probabilities’ or a reliance on hypothetical inﬁnite reference sequences? 22 If ‘propensities’ are primitive, can a propensity theory give us any insight into the nature of probability? 23 How are interpretations of quantum mechanics relevant to philosophical disputes about probability? 24 Can we give up common sense in favour of a ‘many-worlds’ reality containing both Schrödinger’s cat alive and Schrödinger’s cat dead? 25 Are teleological explanations an alternative to causal explanations or a kind of causal explanation? 26 How can we determine whether all sciences form a hierarchy in which the higher can always be reduced to the lower? 27 Do different areas of science describe different realities, or just one physical reality at different levels of detail? 316 10 Philosophy of Biology E L L I O T T S O B E R Philosophy of biology is a branch of the PHILOSOPHY OF SCIENCE (chapter 9). Some of its characteristic questions concern the relationship of biology to the rest of science; others are internal to the methods and results of biology itself. Philosophy of biology also bears a diverse set of relationships to other areas of philosophy. Questions about vitalism and reductionism closely parallel discussions of the mind–body problem in PHILOSOPHY OF MIND (chapter 5). Evolutionary theory’s relationship to the argument from design brings it into contact with the PHILOSOPHY OF RELIGION (chapter 15). And the idea that human evolution has implications about morality and human nature connects it with ETHICS (chapter 6). 1 The Subject Matter of Biology 1.1 The deﬁnition of life Biology is conventionally deﬁned as the science of life. Philosophers, and scientists when they wax philosophical, wonder what being alive amounts to. Should life be deﬁned in terms of the molecule DNA? Well, we know that organisms pass traits to their offspring by transmitting genes to them, and genes are made of DNA. These offspring begin life as one-celled organisms; the genes they contain and the environments they occupy then lead them to grow into multi-cellular organisms containing many different types of cell. Thus, DNA is fundamental to the processes that comprise heredity and development. DNA and RNA are central to organic processes as they occur on earth. But must life be based on DNA and RNA? Could other molecules play the same role in heredity and development? What we earthlings call ‘organic chemistry’ is the chemistry of carbon compounds, but could life forms, if they exist in other galaxies, be built out of silicon? Perhaps it is parochial to think of life in terms of the physical structures that happen to mediate life processes on earth. Is it possible to provide more general and more abstract criteria for what it takes to be alive? As already mentioned, organisms reproduce, pass traits to their offspring, and develop in the course of their lifetimes. In addition, they extract energy from their envi- ronment and use this energy to repair tissue damage and to engage in homoeostatic ELLIOTT SOBER processes that keep some of their characteristics constant despite the ﬂux that occurs	Blackwell
in their environments. Not all living things do all of these, but most do most. Perhaps ‘alive’ is a cluster concept of the sort described by WITTGENSTEIN (chapter 39); being alive means that the system engages in one or more of these characteristic life processes. We best conceive of these life processes in a way that leaves open how many types of physical structure can instantiate them. Although biology is the science of life, being alive is not a theoretical concept in biology. Biology has developed as a discipline without having anything terribly precise inquiry is. ARISTOTLE (chapter 23) once to say about exactly what its domain of observed that it is a mistake to demand more precision of a concept than is needed. Perhaps ‘alive’ is not a crisply delimited category in nature (for parallel remarks about the notion of ‘mind’, see Sober 1990). 1.2 Vitalism, supervenience and reductionism How are the biological properties and processes just listed related to physical properties and processes? This question precisely parallels the mind–body problem in the philoso- phy of mind. Just as Cartesian dualism maintains that individuals have mental proper- ties in virtue of possessing minds that are made of a non-physical substance, so there is a position in philosophy of biology that holds that living things have biological pro- perties in virtue of containing a non-physical substance (a Bergsonian elan vital) that animates them with life. This is vitalism, in one of the many senses that that term has acquired. If vitalism were true, the SUPERVENIENCE THESIS (pp. 80–1) much discussed in the philosophy of mind would be false. This thesis maintains that if two individuals were physically identical and lived in physically identical environments, then they would be identical in all respects. They would have the same psychological properties, and they would have the same biological characteristics as well. The slogan that sums up super- venience is ‘no difference without a physical difference’. Supervenience is a synchronic claim, different from the diachronic claim that constitutes the thesis of causal determinism. Most biologists and philosophers of biology believe that the supervenience thesis is correct. For example, if two organisms are physically identical and live in physically identical environments, then they must have the same ﬁtness value (Rosenberg 1978, 1985; Mills and Beatty 1979; Brandon 1978; Sober 1984). An organism’s ﬁtness is its ability to survive and reproduce. In part, the supervenience thesis has been accepted because the alternative seems never to have led to anything substantial in science. Science has made great strides in understanding the physical bases of respiration, digestion, reproduction and other biological processes. But the hypothesis that living things possess an immaterial something has led to nothing. Still, the question may be asked of how strongly the track records of physicalism on the one hand and vitalism on the other establish that the supervenience thesis is true (Sober 1999a). Although physicalism in the form of the supervenience thesis is widely accepted, most philosophers of biology reject the claim that biology reduces to physics. Their reasons recapitulate the discussion of FUNCTIONALISM (pp. 178–9) and multiple realizability in philosophy of mind. Although an organism’s ﬁtness supervenes on its 318 PHILOSOPHY OF BIOLOGY physical properties and the physical properties of its environment, ﬁtness is not itself a physical property. What do a ﬁt orchid and a ﬁt otter have in common? There is no physical property, or set of physical properties, that deﬁnes what ﬁtness is. Fitness is multiply realizable. The theory of natural selection states generalizations in terms of the concept of ﬁtness; since ﬁtness is not a physical property, the theory of natural selection	Blackwell
cannot be reduced to physics. This is anti-reductionism without vitalism. There are no immaterial substances in the subject matter of biology, but biological properties describe similarities that exist in spite of physical differences. Every ﬁt organism is a physical thing, but ﬁtness is not a physical property. Similar reasoning has led some philosophers to argue that Mendelian genetics cannot be reduced to theories in molecular biology (Kitcher 1984); however, there is room to explore what these theories amount to, and also to question how reductionism should be understood (Waters 1990; Sober 1999b). The multiple realizability argument against reductionism presupposes a very speciﬁc conception of what reductionism requires. However, there are a number of contend- ing philosophical analyses of reduction that need to be evaluated to decide how that concept should be understood. Nagel (1961) set the agenda for subsequent discussion by arguing that one theory reduces to another if the ﬁrst can be deduced from the second, once appropriate ‘bridge laws’ are provided that connect the vocabularies of the different theories to each other. Schaffner (1976) observed that false theories are sometimes reduced to true ones by showing that the reduced theory is a good approxi- mation of the reducing theory (for example, Newtonian mechanics is sometimes said to reduce to the theory of special relativity on the grounds that relativistic particles approach Newtonian trajectories as they slow down); however, this poses a problem for Nagel’s account, since a false theory cannot be deduced from a true one if the bridge laws are true. Schaffner suggested that the reducing theory must correct the reduced theory, and that the two theories must exhibit an appropriate analogy. Other proposals have been made and it has been recognized that ‘reduction’ and ‘reductionism’ each have multiple meanings (Wimsatt 1979; Dupré 1993; Rosenberg 1994). 1.3 Teleology Philosophers of biology have for a long time debated how the concept of teleology (goal- directedness) is treated in modern biology. If it forms a part of that science, does this count against reductionism? After all, it seems clear that teleology has had no place in physics since the seventeenth century. The answer is that philosophers have endorsed various reductionist accounts of what function claims mean, but that the reduction involved is not to physics, but to the general concept of causality. Since causal concepts are widely used in the rest of science, we may draw the conclusion that there is nothing irreducible about teleological discourse in biology. Most philosophical discussion of this issue derives from the work of Wright (1973, 1976) and Cummins (1975). Wright defended an etiological account of function. To describe the function of the heart is to say why organisms have hearts. Hearts evolved in the lineages leading to modern organisms because hearts helped organisms to circulate their blood. The organ was not retained because it made noise. The function of the heart, therefore, is to pump blood, not to make noise. This is not to deny that organisms sometimes beneﬁt from the fact that hearts make noise; for example, babies 319 ELLIOTT SOBER are comforted by hearing their mothers’ heartbeat, and patients are diagnosed when physicians listen to their heartbeats. But, according to Wright, these beneﬁts are not part of the heart’s function if they did not help cause the heart to evolve. Wright’s account crisply distinguishes function from fortuitous beneﬁt. The etiological account faces some interesting objections. Scientists were making function claims long before evolutionary theory was developed. What could Harvey have meant by his claim that the function of the heart is to pump blood? In addition, the Wrightian pattern is sometimes present in traits that do not have a function. If a man fails to exercise because he is obese, and he remains obese because he fails to exer-	Blackwell
cise, are we prepared to say that the function of obesity is to prevent exercise? For these and other reasons, non-etiological accounts of function are of interest (Boorse 1976). Cummins’s proposal was to understand function claims as summarizing a kind of analytic decomposition. If the organism has certain capacities, then the function of its organs may be understood in terms of how they contribute to those capacities. To say that the function of the heart is to pump blood is to describe what hearts now do, not why they evolved. Function claims are like descriptions of the workings of an assem- bly line; the whole factory is able to produce a product because the various stations along the assembly line make their separate contributions to that end result. Cummins’s proposal has been criticized for its liberality. According to Cummins, the heart has as many functions as it has effects on the containing system. The heart pumps blood, but it also makes noise, and it weighs a couple of pounds. Each of these constitutes a func- tion that the heart has, because each contributes to some property of the organism as a whole. Cummins pointed out that we may not be equally interested in all of these effects, which is why we may be disinclined to talk about the function of the heart’s noisiness or weight. However, for Cummins, there is no objective, interest-independent distinction between function and fortuitous beneﬁt. Wright’s etiological account and Cummins’s alternative have been reﬁned, and the debate continues (see Allen, Bekoff and Lauder 1998; Buller 1999). However, it is worth considering the possibility that both concepts are needed in biology (Sober 1993; Godfrey-Smith 1993, 1994). It is important to distinguish the reason a trait evolved from the beneﬁcial effects the trait has once it is present. But it also is important to analyse the workings of the current organism. Which account captures the real meaning of the word ‘function’ may be less important than the fact that both are, broadly speaking, causal accounts. Wright focuses on phylogeny, whereas Cummins focuses on ontogeny. Again, we must realize that ‘function’ is not a theoretical term used in biology, but is an informal concept that is used to talk about biological issues. Clarity is important if we are to avoid miscommunication, but clarity does not always require univocity. 2 The Structure of Evolutionary Theory Although philosophers generally agree with Dobzhansky (1973) and other biologists that evolution is the central unifying idea in biology, a good deal of work goes on in biology that is not evolutionary in its content. Evolution is standardly deﬁned as changes in the genetic composition of populations. But ecologists, physiologists, anatomists and molecular biologists often look at properties and processes found in 320 PHILOSOPHY OF BIOLOGY populations, organisms and organic molecules that do not involve genetic change. Some philosophical work has been done on these non-evolutionary subjects, but the fact remains that evolutionary theory has been the central focus of philosophy of biology. 2.1 Darwin’s two-part hypothesis For a long time, philosophy of biology languished in the shadow of philosophy of physics. In part this was because philosophy of science for many years was developed by philosophers who knew about physics, but often knew little about other sciences; in addition, these philosophers often viewed physical theories like Newtonian mechanics, relativity theory and quantum mechanics as paradigms. Questions about the philoso- phy of science were posed and answered with these physical theories in mind; the net result was that other sciences had to measure up to these physical standards, or be viewed as second rate. This ‘physics worship’ was not just a consequence of the personal inclinations of the philosophers involved; it also was grounded in the philosophical	Blackwell
theses that these philosophers often defended. Reductionism and claims about the unity of science (Oppenheim and Putnam 1958) conferred upon physics a privileged status. If relativity theory and quantum mechanics are one’s paradigms, then it makes sense to deﬁne science as the search for laws of nature. Laws are usually taken to be universal generalizations that do not mention any place, time or individual, and which are empirical and nomologically necessary. Newton’s universal law of gravitation, for example, does not mention any individual place, time or thing; it describes the gravitational attraction that must exist between any two objects that have mass. If science is the search for laws, the question immediately arises of how Darwin’s theory of evolution can count as scientiﬁc. It is difﬁcult to ﬁnd a law of evolution in Darwin’s writings. His major insight was to formulate, test and apply the following two hypotheses: (1) all the organisms now on earth share common ancestors – there is a single tree of life; (2) natural selection was an important cause of the characteristics that living things on earth are observed to have. Neither of these claims is a law; both are historical hypotheses about the earth’s past. Nomothetic Sciences and Historical Sciences Rather than consign the claims for the single tree of life and natural selection to the back- ground of Darwin’s achievement and promote some universal claim to the centre of our attention, it is better to recognize that not all sciences have the search for laws as their primary goal. Nomothetic sciences are like this, but historical sciences are not. The con- trast in physics between mechanics and astronomy illustrates this difference. In mechan- ics the main goal is to discover the laws of motion. One may want to discover the truth values of singular statements about particular objects in order to test proposed laws, but this information about particulars is merely a means to an end. In astronomy the main goal is to discover facts about the composition and history of objects in the cosmos. One may use laws of nature to test singular historical hypotheses, but this information about laws is merely a means to an end. Darwin’s theory of evolution is ﬁrst and foremost a historical hypothesis; it is an attempt to reconstruct the history of the living things we ﬁnd around us. Darwin’s theory of evolution and Newton’s theory of gravitation are different, not just in their subject matters, but in their logical structure. 321 ELLIOTT SOBER Sparrows Robins Crocodiles Sparrows Robins Crocodiles (SR)C S(RC) Figure 10.1 2.2 Phylogenetic inference Darwin argued that the best explanation of many of the similarities we observe among different species is the hypothesis that they share common ancestors. However, not all similarities have this evidential signiﬁcance; for example, the fact that sharks and dolphins both are shaped like torpedoes can easily be explained by the hypothesis that this shape was useful for each group of organism to move efﬁciently through water. We would expect to ﬁnd the torpedo shape even if the two groups had arisen independently. In contrast, those similarities that cannot be explained by the functional requirements of similar environments provide better evidence of common descent. This is why vestigial organs are such telling indications of shared ancestry. Why do human foetuses have gill slits, if they are not related to organisms in which gill slits are useful? The logic of these inferences conforms to the Likelihood Principle: a set of observations is said to favour one hypothesis over another precisely when the ﬁrst hypothesis confers on the observations a higher probability than the second one does (Sober 1988, 1993, 1999c; Royall 1997). A similarity S exhibited by two species is evidence that they have a common ancestor precisely when Pr (S/common ancestry) > Pr (S/separate ancestry).	Blackwell
If all species on earth share common ancestors, the question remains as to which species are closely related and which are related only more distantly. The methodology of this problem has been intensively studied in evolutionary biology. One principle that is often used is termed phylogenetic parsimony. It was independently proposed by Hennig (1966) and by Edwards and Cavalli-Sforza (1964). Consider the two phylogenetic trees depicted in ﬁgure 10.1. The tips of the trees represent present-day species or superspe- ciﬁc taxa; interior nodes represent ancestors. The (SR)C tree represents the hypothesis that sparrows and robins have a common ancestor that is not an ancestor of croco- diles; the S(RC) tree represents the claim that robins and crocs are more closely related to each other than either is to sparrows. We know by observation that sparrows and robins have wings, but crocodiles do not. Can this observation be used to discriminate between the two genealogical hypotheses? If we assume that the species at the root of the tree lacked wings, then we are saying that sparrows and robins share a derived character (an apomorphy) that crocodiles do not possess. The (SR)C tree is more parsimonious than the S(RC) tree in this case, because (SR)C is able to explain the observations at the tips of the tree by 322 PHILOSOPHY OF BIOLOGY Ingroup Outgroups wingless wingless ? Figure 10.2 postulating a single change in character state (from wingless to winged) in the tree’s interior, whereas S(RC) requires two such changes to explain the observations. On the other hand, if the species at the root of the trees had wings, then sparrows and robins would share an ancestral character (a plesiomorphy), and the two hypotheses would be equally parsimonious, since each could explain the observations by postulat- ing a single change in character state. Thus, the principle of phylogenetic parsimony asserts that derived similarities are evidence of relatedness, while ancestral similarities are not. How is one to determine whether wings or lack of wings is the ancestral condition in the problem just described? The principle of parsimony addresses this question as well. Even if we don’t know how robins, sparrows and crocodiles are related, we still may know that these three groups are more closely related to each other than any of them is to various other groups (for example, daffodils). If the outgroup species depicted in ﬁgure 10.2 all lack wings, then phylogenetic parsimony dictates that we should conclude that winglessness is the ancestral condition for the ingroup as well. To infer phylogenetic relationships, why should parsimony be used, rather than a criterion of overall similarity, which accords evidential signiﬁcance to derived and ancestral similarities alike? Some have argued that Popperian considerations con- cerning FALSIFIABILITY (pp. 287–90) sufﬁce to justify phylogenetic parsimony (see, for example, Eldredge and Cracraft 1980; Wiley 1981); others have suggested a likelihood analysis instead (Sober 1988). Phylogenetic inference provides an interesting context for considering philosophical questions about the role of a parsimony principle in scientiﬁc reasoning. Although philosophers often complain that it is unclear what it means for one theory to be simpler or more parsimonious than another, it is clear enough what it means for one geneal- ogical hypothesis to provide a more parsimonious explanation of the observations than another. And in answer to the question of why we should use parsimony to infer genealogical relatedness, it seems unsatisfactory to reply that a preference for parsi- monious hypotheses is part of what it means to think scientiﬁcally. Why is it ‘unscien- tiﬁc’ to use overall similarity as one’s guide? Parsimony is not an end in itself; if a 323 ELLIOTT SOBER preference for parsimonious explanations is justiﬁed, it is justiﬁed because it provides a	Blackwell
means of achieving some more ultimate epistemic end. 2.3 Adaptationism Natural selection is one among several processes that can inﬂuence the evolution of a trait. Since a lineage starts evolving with some pre-existing set of ancestral traits, the characteristics exhibited by descendants may show the inﬂuence of that ancestral condition. For example, suppose an ancestral bear species has a fur thickness of 5 cm. The climate then gets colder, so that the optimal fur thickness for one of its descendants would be 12 cm. If the descendant achieves a fur thickness of, say, 10 cm, one may want to attribute this outcome both to natural selection and to phylogenetic inertia, sometimes called ancestral inﬂuence (Orzack and Sober 2001). Another factor that can prevent natural selection from moving species to optimal trait values is the underlying genetics. For example, if the ﬁttest phenotype is coded by a heterozygote, it will be impossible for all individuals to exhibit that optimal phenotype. In similar fashion, random genetic drift, correlation of characters, and other factors can prevent natural selection from causing the ﬁttest available phenotype to evolve (Sober 1993). The debate about ‘adaptationism’ that has been going on for the last twenty-some years in evolutionary biology does not concern Darwin’s tree of life hypothesis; nor does it concern the modest claim that natural selection has inﬂuenced most of the pheno- typic traits that species exhibit. Rather, the question concerns the power of natural selection. Is selection merely one among several important inﬂuences on trait evolution, or is it the only important inﬂuence (Orzack and Sober 1994; Sober 1993)? Can phylogenetic inertia and genetic constraints be ignored in predicting the descendant bear’s fur thickness? Adaptationism embodies a relatively monistic conception of trait evolution; its alternative is evolutionary pluralism (Gould and Lewontin 1979). Adaptationism, Pluralism and Methodology Separate from the substantive debate about the processes that have governed evolution, the debate about adaptationism has had an important methodological dimension. Gould and Lewontin (1979) accused adaptationists of endorsing ‘just-so stories’ – of accepting explanations uncritically. Adaptive hypotheses need to be tested rigorously, which means that they should be tested against non-adaptive alternative explanations. Gould and Lewontin also suggested that adaptationism violates Popper’s injunction that scientiﬁc hypotheses should be falsiﬁable. Notice that neither of these criticisms, even if they were correct, would show that pluralism is true and monism is false as claims about nature. The fact that a proposition has been accepted uncritically does not entail that the proposition is false. The charge that adaptationism is untestable should not be taken at face value. It is true that if one adaptationist hypothesis is rejected, then another can be invented. This is because adaptationism is an ism – it speciﬁes the kind of explanation that all or most traits have without saying anything in detail about what speciﬁc adaptationist hypothe- ses are true. Consider, for example, the fact some species reproduce sexually while others 324 PHILOSOPHY OF BIOLOGY reproduce asexually. Adaptationists will insist that this variation exists because sexual reproduction is better in some circumstances and asexual reproduction is better in others. However, this general claim does not tell us why selection favours sexual repro- duction in some cases and asexual reproduction in others. Adaptationism is a research programme in something like the sense that Lakatos (1978) described (Mitchell and Valone 1990); the failure of one or several adaptationist models does not mean that the research programme is bankrupt. The same point applies to evolutionary pluralism: if one pluralistic model fails, another can be invented, but this does not show that plural-	Blackwell
ism should be rejected. It is important that scientists rigorously test speciﬁc hypotheses, whether they are adaptationist or pluralistic in form; however, the isms that govern research programmes are not so easily tested. Only the accumulation of successes and failures over the long run will tell us whether adaptationism is true as a claim about nature (Orzack and Sober 1994). 2.4 The units of selection controversy Whether or not the concept of function is understood historically (section 1.3), it is pretty clear that evolutionists use the concept of adaptation in this way. To say that wings are an adaptation for ﬂying does not mean that wings now are useful because they help organisms to ﬂy; it means that wings evolved because they helped organisms to ﬂy. The relevant consideration is in the past tense, not the present. Wings could be an adaptation for ﬂying even if it now is disadvantageous for the organisms in question to leave the ground. Historical origin and present advantage are not automatically linked. Notice also that wings can be adaptations for ﬂying even if adaptationism is wrong in what it says about wings; to say that natural selection helped cause the trait to evolve does not mean that selection was the only important inﬂuence on the trait’s evolution. When Darwin argued that this or that trait is an adaptation, he usually had in mind that the trait evolved because it helped individual organisms to survive and reproduce. For example, predators have sharp teeth because individuals with sharp teeth do better in the struggle for existence than individuals that have dull teeth. The explanation is not that the trait evolved because it helped the whole species, or the whole ecosystem. The adaptations that Darwin discussed are mainly individual adaptations. These indi- vidual adaptations evolved by the process of individual selection, wherein individuals compete with other individuals in the same species. There were a few occasions, however, in which Darwin saw things differently. Why are there sterile workers in some species of social insect? This is not because sterile indi- viduals do better in surviving and reproducing than fertile individuals. Rather, Darwin argued that nests that contain sterile workers do better than nests that do not. Darwin introduced the idea that some traits are group adaptations – they evolved because they helped groups, not because they helped individual organisms. These adaptations evolved by the process of group selection, wherein groups compete with other groups in the same species. Darwin thought that group selection is required to understand important features of human morality. Here is how he analysed the problem in The Descent of Man: 325 ELLIOTT SOBER It must not be forgotten that although a high standard of morality gives but a slight or no advantage to each individual man and his children over the other men of the same tribe, yet that an increase in the number of well-endowed men and advancement in the stan- dard of morality will certainly give an immense advantage to one tribe over another. There can be no doubt that a tribe including many members who, from possessing in a high degree the spirit of patriotism, ﬁdelity, obedience, courage, and sympathy, were always ready to aid one another, and to sacriﬁce themselves for the common good, would be victorious over most other tribes; and this would be natural selection. At all times throughout the world tribes have supplanted other tribes; and as morality is one important element in their success, the standard of morality and the number of well-endowed men will thus everywhere tend to rise and increase. (Darwin 1981: 166) The character traits that Darwin describes are disadvantageous to individuals, but advantageous to groups. A courageous man who risks his life in warfare is less ﬁt than a coward in the same group who plays it safe; yet groups that include courageous	Blackwell
individuals are ﬁtter than groups that include only cowards. Biologists came to use the term ‘altruistic’ for traits that are deleterious to indivi- duals but advantageous to groups. This terminology is potentially confusing, since altruism, so deﬁned, describes the ﬁtness costs and beneﬁts of the trait, not how or even whether the individual thinks and feels. Sterile workers in a nest of social insects and the warriors whom Darwin described are both altruistic in the evolutionary sense, even though the warriors have views about their own behaviour while the insects presum- ably do not. And just as a mindless organism can be an evolutionary altruist, it also is possible for a behaviour that enhances the actor’s own ﬁtness to be motivated by desires that are psychologically altruistic (Sober and Wilson 1998). The idea that evolution involves both individual selection and group selection was standard fare in the heyday of the Modern Synthesis in evolutionary biology, roughly from 1930–60. Biologists invoked individual selection to explain some traits and group selection to explain others. Mathematical population geneticists, such as R. A. Fisher, J. B. S. Haldane and Sewall Wright, brieﬂy expressed reservations about the ability of group selection to cause altruistic traits to evolve, but these theoretical points were not given much credence by biologists in the trenches; these biologists thought they observed altruism in nature and they sought to explain altruism in the only way they knew how – by invoking the hypothesis of group selection. This complacent pluralism about the units of selection was shattered in the 1960s, when group selection was attacked by a number of biologists. The most thorough and devastating critique was G. C. Williams’s (1966) book Adaptation and Natural Selection. Williams touched a nerve and his vigorous rejection of adaptations that exist for the good of the group spread quickly through the evolution community. The arguments that Williams advanced were popularized ten years later by Richard Dawkins (1976) in his widely read book The Selﬁsh Gene. The idea that some traits evolve because they beneﬁt the group was replaced by genic selectionism – the idea that all traits evolve because they are good for the gene. However, not all biologists subscribed to this new viewpoint; they argued that group selection had been rejected prematurely (for example, Lewontin 1970; Wilson 1980; Wade 1978). Since then, multi-level selection theory – the view that selection occurs at all levels of the biological hierarchy – has been making a comeback (Sober and Wilson 1998). 326 PHILOSOPHY OF BIOLOGY Assessing Group Selection: An Empirical or Conceptual Controversy This scientiﬁc controversy is philosophically interesting because group selection was rejected in the 1960s for a number of reasons, and these reasons differ so much in character that it is unclear what in the controversy is empirical and what is conceptual. The schizophrenic character of the arguments that biologists produced may be illustrated by their discussion of the evolution of sex ratio – the mix of males and females that a population exhibits. Williams (1966) treats this as an empirical test case for group selection. He notes, following Fisher (1957), that individual selection would favour an even sex ratio, whereas group selection would sometimes favour a female-biased sex ratio (the latter because groups maximize their productivity by having the smallest number of males consistent with all the females being fertilized). Williams claims that the sex ratios observed in nature are uniformly even, and concludes that the empirical hypothesis of group selection is mistaken, at least as it pertains to the evolution of sex ratio. A year later, W. D. Hamilton published a paper called ‘Extraordinary Sex Ratios’. Hamilton (1967) reported that female-biased sex ratios are very common in the social	Blackwell
insects. One might expect that this empirical ﬁnding, interpreted through the lens of Williams’s argument, would have led biologists to conclude that group selection is impli- cated in the evolution of sex ratio. However, this is precisely how the model was not interpreted. The reason is that Hamilton’s paper provides a mathematical theory for understanding how uneven sex ratios evolve. Solving Hamilton’s equations allows one to identify the ﬁttest (the ‘unbeatable’) sex ratio strategy that an individual female should follow in determining her mix of sons and daughters; this formalism was widely inter- preted as showing how individual selection could promote the evolution of a female- biased sex ratio. One of the few biologists who did not interpret Hamilton’s model in this way was Hamilton himself; Hamilton says, though only in a footnote, that his model represents the action of group selection. Although Hamilton embraced the idea of group selection in 1967 and later publications, his followers were more Hamiltonian than Hamilton himself. Williams (1992) concurs with Hamilton’s interpretation. Yet many biologists con- tinue to view group selection as beyond the pale, and they cite Hamilton and Williams as having shown why. Sterelny and Kitcher (1988) argue in favour of genic selectionism on the grounds that all adaptations can be interpreted as evolving because they beneﬁt the genes that code for them. Biased sex ratios, extreme altruism and so on, can all be brought under the rubric of the selﬁsh gene. The reason group selection is the wrong way to think about nature is not that it is empirically disconﬁrmed; rather, the problem is that it is insufﬁ- ciently general. Although all adaptations can be treated as genic adaptations, it is false that all adaptations can be interpreted as group adaptations. According to Sterelny and Kitcher, the solution to the units of selection problem involves adopting a convention, not answering a factual question about the history of life. For them, the fact that a trait can be interpreted as a genic adaptation, no matter what the trait is like, is a point in favour of the genic convention. However, those who think that the units of selection problem is an empirical issue view this as a defect, not a strength, of the framework that Sterelny and Kitcher recommend (Wimsatt 1980; Sober and Lewontin 1982; Lloyd 1988; Sober and Wilson 1994, 1998). 327 ELLIOTT SOBER 2.5 Are there laws of evolution? Although Darwin’s fundamental contribution was the formulation, test, and applica- tion of two historical hypotheses (see also Kitcher 1993: ch. 2), the broad subject of modern evolutionary theory includes much more than historical hypotheses. The subject is peppered with what biologists call ‘models’. These are mathematical statements that describe what the evolutionary consequences would be of speciﬁed initial conditions. For example, Fisher’s analysis of sex ratio is an if/then statement. Fisher (1957) described a sufﬁcient condition that ensures that a population with an uneven sex ratio will evolve towards an even sex ratio, and remain there. Fisher’s theory has all the earmarks of a law of nature, save one. It is general, it does not refer to any place, time or individual, and it has counterfactual force. However, Fisher’s model is an a priori mathematical truth. This does not mean that the claim that Fisher’s model applies to a real world population is a priori; whether the population exhibits random mating and satisﬁes the other assumptions of the model is an empirical matter, and whether the population exhibits an even sex ratio is an empirical matter as well. The antecedent is empirical and the consequent is too, but the conditional is a priori (Sober 1984, 1993). Empirical generalizations exist in evolutionary biology, but they often fail to have the	Blackwell
modal force that laws are supposed to have. Cope’s Rule, for example, says that species evolve in the direction of increased size. This is, at best, a rule of thumb that summa- rizes a pattern found in the fossil record (Hull 1974). The problem is not just that there are exceptions (a law, after all, can be probabilistic), but that there is nothing in our understanding of the evolutionary process that tells us that evolution must have this sort of directionality. A fundamental property of natural selection is that it is ‘oppor- tunistic’. If growing larger is advantageous, selection will push the population along that trajectory; but if growing smaller is advantageous, selection will favour that outcome, instead. It is an accident of circumstance whether the environment favours one transition rather than the other. If the history of life on earth happens to contain more environments of the ﬁrst kind than of the second, Cope’s Rule will be true as a statistical generalization, but it will not be a law. The same point holds about increasing complexity. To be sure, complexity has often increased; however, there also are many cases in which complexity has declined. For example, the evolution of parasites from their free-living ancestors often involves the loss of organs; selection often leads organisms to lose features that they no longer need. Still, if life started simple, an increase in complexity was bound to occur. Biologists gen- erally regard this as an artefact of the initial conditions that obtained, not as reﬂecting some resolute directionality in the evolutionary process itself. As an analogy, consider a game played on a one-row checker board that has a thousand squares. The game is played in a sequence of steps; at each step, a coin is tossed to decide how the checker will be moved. Heads means the checker moves one square to the right and tails means that it moves one square to the left. Of course, if the coin lands tails when the checker is on the left-most square, the checker can’t move to the left; in this case, one must toss the coin again. The same applies if the checker is on the right-most square when the coin lands heads. Since the coin is fair, there is no directionality in the laws that govern the checker’s trajectory. The laws entail no net tendency for the checker to move to the 328 PHILOSOPHY OF BIOLOGY right or to the left. However, an asymmetry can be introduced by the checker’s initial location. Suppose we begin the game with the checker in the left-most square. After ﬁfty tosses, the checker will almost certainly occupy a square that is to the right of where it began (Maynard-Smith 1988; Gould 1989; Sober 1994a). I so far have suggested that there are two types of generalization in evolutionary biology that fail to be scientiﬁc laws in the usual sense. The conditional statements that comprise mathematical models are a priori, and empirical generalizations about evolu- tionary trends often are accidentally true, if true at all. This does not show that there are no laws. Are there any? Beatty (1995) has argued that there are none, not just none in evolutionary biology, but none in biology as a whole. His reason for this claim is that every biological generalization is true in virtue of a contingent evolutionary process’s making it so. For example, if the organisms in a population obey Mendel’s law of inde- pendent assortment, this is because evolution led this regularity to evolve. If selection had favoured some other pattern of heredity, that pattern might have evolved instead. This is Beatty’s evolutionary contingency thesis. One question that needs to be considered about Beatty’s argument is depicted in the following diagram: ﬁ E t 0 (If P then Q) t t 1 2 If the evolutionary regularity ‘If P then Q’ holds true between times t1 and t2 only because contingent evolutionary events E happened to take place at time t0, then it makes sense to say that the regularity is contingent. However, this leaves it open that	Blackwell
the more complex conditional ‘if E occurs, then (if P then Q) will be true later’ holds true non-contingently (Sober 1997a). This point does not establish that biological laws exist, but it does show that one cannot establish that there are no laws just by pointing out that regularities depend on earlier contingencies. Furthermore, if causality entails the existence of laws (a metaphysical claim that should not be accepted uncritically; Anscombe (1975), for example, denies it), then the causal dependency of ‘If P then Q’ on E entails the existence of a law. Sensitivity to Initial Conditions Gould (1989) has emphasized the idea that evolutionary outcomes often exhibit a great deal of sensitivity to initial conditions. Even a small change in the conditions that obtained millions of years ago would have made a huge difference in the distribution of life forms that we ﬁnd on earth now. If a certain meteor strike had not occurred, the dinosaurs would not have gone extinct; if the dinosaurs had not gone extinct, mammals would not have proliferated; and if the mammalian explosion had not occurred, no mammal resem- bling Homo sapiens would have made its appearance. According to Gould, ‘replaying the tape’ would reveal that many of the interesting features of the life forms we ﬁnd around us are accidents of history. The claim that there is sensitivity to initial conditions has the following form: although X1 leads to Y1, it also is true that X2 leads to Y2, where X1 and X2 are very similar, while Y1 and Y2 are very different. The usual gloss on this claim is that there are two laws: 329 ELLIOTT SOBER X1 leads to Y1 and X2 leads to Y2. Thus, the sensitivity thesis does not entail that there are no biological laws. Notice also that Gould is not claiming that all features of living things exhibit sensitivity to initial conditions; he is merely calling attention to the importance of this phenomenon. It is perfectly possible that some features of life are robust (that is, insensitive to perturbations in initial conditions) while others are not. The frequency of evolutionary convergence shows that similar outcomes often arise from rather different starting points. What is sorely needed in this area are biologically well motivated theo- ries concerning which features should be robust and which should exhibit sensitivity. Evolutionary biology very much needs to move this problem beyond the reporting of piecemeal intuitions about examples. 3 The Philosophical Signiﬁcance of Evolutionary Theory 3.1 The argument from design The two parts of Darwin’s theory of evolution by natural selection both conﬂict with at least one traditional reading of Genesis. First, the theory postulates a single tree of life, and thus contradicts the claim that different groups of organisms were separately created. Second, the theory postulates the mindless process of natural selection as the cause of life’s adaptive features, and thus contradicts the claim that adaptive features arose by the intelligent design of a creator. There are additional, though less central, points of conﬂict as well; for example, Darwin held that the earth is ancient while one literal reading of the Bible says that the earth is young. Not only does the theory of evolution conﬂict with one set of religious doctrines; it also conﬂicts with an inﬂuential argument for the existence of God – the ARGUMENT FROM DESIGN (pp. 478–80). The ﬁfth of THOMAS AQUINAS’s (chapter 24) ﬁve ways of proving the existence of God contends that things that ‘act for an end’ must either have minds or be produced by an intelligent designer. In the century in Britain preceding Darwin’s publication of the Origin, this argument was repeatedly elaborated. It was not simply one argument for the existence of God; it became the fundamental argument. Perhaps the most famous formulation of the design argument is that given by William Paley, in his book Natural Theology (1805). Paley proposed an analogy. If you	Blackwell
were walking across a heath and found a stone, you would not dream of concluding that the stone was produced by an intelligent designer. However, if you found a watch, you would conclude without hesitation that the watch was produced by a watchmaker. The difference between the watch and the stone is that the watch exhibits adaptive complexity. The watch as a whole measures out equal intervals of time, and the parts of the watch conspire to allow the watch to perform that function; were any of those parts even modestly different from the way they are, the watch would not keep time. Paley then argues that what is true of the watch also holds for many features of organisms. The vertebrate eye, for example, also exhibits adaptive complexity. Paley concludes that organisms are the result of intelligent design no less than the watch is the handiwork of a watchmaker. Philosophers have not always agreed on what the form of the design argument is. For example, HUME (chapter 31), in his Dialogues Concerning Natural Religion (1779), 330 published some thirty years before Paley’s Natural Theology appeared, suggests that the design argument is an analogical argument that has the following form: PHILOSOPHY OF BIOLOGY Watches are produced by intelligent design. Organisms are like watches. Organisms were produced by intelligent design. The double line separating premisses from conclusion indicates that the argument is not intended to be deductively valid. Hume suggests that the strength of the analogi- cal argument depends on the degree to which watches and organisms resemble each other – the more similar they are, the more strongly the premisses support the conclu- sion. He then points out that there are many dissimilarities – watches are made of metal and glass, organisms breathe, and so on. Hume concludes that the design argument is a very weak analogical argument. A more charitable reading of the argument from design construes it as an inference to the best explanation (pp. 42–3) (what C. S. PEIRCE (chapter 36) called an abductive argu- ment), one that does not depend on there being any strong degree of overall similarity between organisms and artefacts. According to this reading of the argument, Paley is using the Likelihood Principle to discriminate between two hypotheses: Od Organisms were created by an intelligent designer. Or Organisms were produced by random mindless processes. Paley claims that the observed adaptive complexity of organisms is rendered much more probable by the former hypothesis than by the latter. Paley discusses the watch on the heath to make vivid the power of this type of inference in an uncontroversial example. The adaptive complexity of the watch favours the ﬁrst of the following hypotheses over the second: Wd The watch was created by an intelligent designer. Wr The watch was produced by random mindless processes. The point of the analogy is just to explain how the Likelihood Principle works. It does not matter how similar or dissimilar watches and organisms are overall, so long as they share the single feature of adaptive complexity (Sober 1993). While he studied at Cambridge, Darwin read Paley and greatly admired Paley’s discussion of adaptive contrivances. Darwin eventually came to reject the argument from design, and to move towards agnosticism, because he thought he saw so many features of organisms that conﬂict with the hypothesis that organisms were created by a benevolent, omniscient and omnipotent designer. For example, Darwin was revolted by the parasitic wasps that lay their eggs on paralysed caterpillars; when the eggs hatch, they eat their hosts alive, leaving the brain until last (Desmond and Moore 1991). If an intelligent designer built so much suffering into the living world, this designer does not deserve our reverence. Darwin saw the living world as saturated with pain and death; this is what one should expect according to the hypothesis of evolution by natural 331	Blackwell
ELLIOTT SOBER selection, but not what one should expect according to the hypothesis of intelligent design as Darwin understood that hypothesis. This is a version of the ARGUMENT FROM EVIL (pp. 480–3). The argument is not best put by saying that the quantity of evil found in the world deductively entails that there is no God; rather, the observations are claimed to play the non-deductive role of favouring atheism over theism. This Darwinian argument is a likelihood argument, just like Paley’s, but Darwin considers a hypothesis that Paley does not, and he considers an observation that differs from the one on which Paley focused. The hypothesis of evolution by natural selection does not claim that living things change their features at random. A process is random when its possible outcomes are equiprobable, or nearly so. Tossing a coin, spinning a roulette wheel, drawing cards from a deck – these are examples of random processes. However, natural selection is a process in which some traits have much higher proba- bilities of spreading than others. The rule that governs selection processes is that ﬁtter traits tend to increase in frequency and less ﬁt traits tend to decline. In addition to elaborating a hypothesis that Paley did not consider, Darwin also emphasized the ubiquity of adaptive imperfections. Vestigial organs, for example, are not what one should expect to ﬁnd if living things were created from scratch by a benevolent, intelligent and powerful engineer. However, since natural selection works on the array of variation that ancestral populations happen to contain, it is no surprise that these processes leave traces of ancestral forms that have no adaptive rationale. Paley anticipated one aspect of this Darwinian argument. He says that we should infer a watchmaker when we see the watch, even if the watch turns out to be imperfect. Paley emphatically denies that imperfect adaptation undercuts the design argument. To be sure, Paley delighted in what he took to be the many perfections found in nature. However, it is clear that he did not think of this as part of his argument for the existence of a designer. Rather, the ostensible perfection of nature comes in later, as evidence that bears on the designer’s characteristics. Adaptive complexity, even when imperfect, proves there is a designer. Further details then show that this designer is benevolent. What, then, is the logical status of the design argument, and how is Darwin’s theory related to it? Darwin and his successors have claimed that imperfect adaptations favour evolutionary theory over intelligent design (see, for example, Gould 1980). Should they concede that perfect adaptation would favour intelligent design over purely random mindless processes? I suggest that this interpretation concedes too much to Paley. Even a perfect watch is no evidence of intelligent design unless we have independently attested auxiliary information about the intentions and abilities that the putative designer would have if he existed. What probability does the design hypothesis confer on the features of the watch that we observe? We observe, for example, that the watch is made of metal and glass. How probable is it that an intelligent designer would use these materials? If intelligent designers never use these materials, the design hypothe- sis will be less likely than the hypothesis of random mindless processes. However, if we assume that intelligent designers often use metal and glass, we reach the opposite verdict. And if we simply do not know what the putative intelligent designer would be inclined to do, we will be unable to compare the likelihoods of the two hypotheses. Paley is right that we do not and should not hesitate to infer a watchmaker from the observed features of the watch; this is because we know a great deal about the incli- nations and abilities that human designers have. However, when we shift from watch 332	Blackwell
PHILOSOPHY OF BIOLOGY to organism, the situation changes. What do we know about the desires and abilities of this putative designer of organisms? I suggest that we do not know enough to compare the likelihoods of the two hypotheses. Paley’s analogy between watch and organism in fact conceals a deep disanalogy. As a result, the design argument fails on its own terms; there is no need to bring in Darwinian theory to see this. It is not true that the observed features of organisms put the design hypothesis to the test and that what we observe disconﬁrms the hypothesis; rather, the present suggestion is that the design hypothesis is untestable (Sober 1993, 1999d). Defenders of the design argument are not entitled to assume without argument propositions that describe the features that organisms would have if they were produced by an intelligent designer. Critics of the argument are not entitled to do this, either. 3.2 Species, essentialism and human nature Species have long been a favourite example that philosophers cite when they discuss natural kinds. For example, MILL (chapter 35) in his System of Logic (1874) claims that human being is a natural kind, but the class of snub-nosed individuals is not, on the grounds that ‘Socrates is a human being’ allows one to predict many other characteristics that Socrates has, but ‘Socrates is snub-nosed’ does not. Aristotelian essentialism burdens the concept of natural kind with a more substantial characterization. Natural kinds not only have predictive richness; in addition, they have ESSENCES (pp. 112–13). The essence of a natural kind is the necessary and sufﬁcient condition that members of the kind must satisfy in order to be members. An individual’s possessing this essence is what makes it belong to the kind in question. In addition, it is a necessary truth that all and only the members of the kind have this essential property. The essence also does a good deal of explanatory work; the fact that an individual has this species-typical essence explains many other features that the individual possesses. Besides citing biological species as examples, philosophers often point to the chemi- cal elements as paradigm natural kinds. Gold is a kind of substance; its essence is said to be the atomic number 79. This atomic number is what makes a lump of matter an instance of gold. And atomic number explains many other properties that gold things have. Kripke (1972) and Putnam (1975) have suggested that science is in the business of empirically discovering the essences of natural kinds. Formulated in this way, essen- tialism is not established by the existence of trivial necessary truths. It is a necessary truth that all human beings are human beings, but this does not entail that there is an essence that human beings have. It is also important to separate the claim that kinds have essences from the claim that individuals in the kind have essential properties (Enç 1986). The fact that gold has one atomic number and lead another does not entail that an individual cannot persist through time as it changes from being made of lead to being made of gold. The example of the chemical elements illustrates a further feature that kind essences are supposed to have. Gold’s essence is purely qualitative; ‘atomic number 79’ does not refer to any place, time or individual. In principle, gold could exist at any place or at any time. What makes two things members of the same natural kind is that they are similar in the requisite respect. There is no requirement that they be causally or spatio-temporally related to each other in any way. 333 ELLIOTT SOBER Although philosophers who accept this essentialist picture of the chemical elements usually think that chemistry has already discovered the essences that various chemi- cal kinds possess, they must admit that biology has to date not delivered the goods with respect to biological species. If biological species have essences and these essences are not beyond the ken of science, then the essentialist must claim that biology will even-	Blackwell
tually reveal what these species essences are. However, there are strong reasons to think that Darwin’s theory of evolution undermines this essentialist picture of biological species (Hull 1965; Mayr 1975; Sober 1980). It is not just that biology has not dis- covered these species essences yet; rather, the way species are conceptualized in evolu- tionary theory suggests that species simply do not have essences. They are not natural kinds, at least not on the usual essentialist construal of what a natural kind is. The reasons for this conclusion need to be stated carefully. The fact that species evolve is, per se, not a conclusive argument against essentialism. Just as the essential- ist can agree that lead might be transmuted into gold, so the essentialist can agree that a lineage might be transformed from one species into another. And the fact that there are vague boundaries between species is not, in itself, a refutation of essentialism, either. If a piece of matter is transformed from lead to gold, perhaps there will be intermediate stages of the process in which it is indeterminate whether the matter belongs to one natural kind or the other (Sober 1980). Unfortunately, there still is disagreement in evolutionary biology about how the species category should be understood. Perhaps the most popular deﬁnition is Mayr’s (1963, 1970) biological species concept (for others, see Ereshefsky 1992). Its anti- essentialist consequences are to a large degree also the consequences that other species concepts have, so we may examine it as an illustrative example. Mayr’s basic idea is that a biological species is an ensemble of local populations that are knit together by gene ﬂow. The individuals within local populations reproduce with each other. And migration between local populations means that there is reproduction between individuals in different populations as well. This system of populations is reproductively isolated from other such systems. Reproductive isolation can be a simple consequence of geographical barriers, or it can mean that the organisms have behavioural or physiological features that prevent them from producing viable fertile offspring even when they are brought together. A consequence of reproductive isolation is that two species can evolve different characteristics in response to the different selection pressures imposed by their different environments. However, the different phenotypes that evolve are not what make the two species two; it is reproductive isolation, not physical dissimilarity, that is deﬁnitive. Mayr (1963) initially allowed two populations to belong to the same species if there was actual or potential interbreeding between them, but he later changed the deﬁni- tion so that actual interbreeding was required (Mayr 1970). This raises the question of what the time scale is on which interbreeding must take place. How often must indi- viduals in different local populations reproduce with each other for the two populations to belong to the same species? Indeed, the same question can be posed about indivi- duals living in the same local population. Another detail that needs to be added to the sketch just provided concerns individuals that exist at different times. Human beings who are alive now are not having babies with human beings who lived thousands of years ago. Reproduction is something that occurs between contemporaneous 334 PHILOSOPHY OF BIOLOGY individuals. So what makes human beings now and human beings thousands of years ago members of the same species? A necessary condition is that human beings now are descended from human beings a thousand years ago. But this is clearly not sufﬁcient; otherwise, a present-day species could not be descended from a different, ancestral, species. Finally, I should note that Mayr’s deﬁnition excludes the possibility of asexual	Blackwell
species, and this is another of its features that has made it controversial. The thing to notice about Mayr’s deﬁnition is that qualitative similarity is neither necessary nor sufﬁcient for conspeciﬁcity. Members of the same species may have very different characteristics. And if creatures just like tigers evolved independently in other galaxies, they would not belong to the species to which earthly tigers belong. What makes for conspeciﬁcity are the causal and historical connections that arise from repro- ductive interactions. Biological species and chemical elements are very different in this regard. Evolutionary biologists talk about species in the same way they discuss individual organisms. Just as individual organisms bear genealogical relationships to each other, so species are genealogically related. Just as individual organisms are born, develop and die, so individual species come into existence, evolve and go extinct. These considerations led Hull (1978, 1988) to develop Ghiselin’s (1974) suggestion that species are individuals, not natural kinds. There is room to doubt, however, that species are as functionally integrated as individual organisms often are. The parts of a tiger depend on each other for survival; excise an arbitrary 30 per cent of a tiger, and the whole tiger dies. However, the extinction of 30 per cent of a species rarely causes the rest of the species to go extinct. This suggests that individuality (in the sense of functional interdependence of parts) comes in degrees, and that species are often less individualistic than organisms often are. Still, Hull and Ghiselin’s main thesis remains; perhaps it should be stated by saying that species are historical entities (Wiley 1981). Similar points apply to higher taxonomic categories. Although ordinary language may suggest that carnivores all eat meat and that mammals all nurse their young, this is not how biologists understand Carnivora and Mammalia as taxa. These taxa are understood genealogically; they are monophyletic groups, meaning that they include an ancestral species and all of its descendants (Sober 1992). Pandas belong to Carnivora because they are descended from other species that belong to Carnivora; the fact that pandas are vegetarians does not matter. Superspeciﬁc taxa, like species themselves, are conceptualized as big physical objects; they are chunks of the genealogical nexus. And just as species are often not very individualistic, superspeciﬁc taxa are even less so (Ereshefsky 1991). The chemical kinds do not comprise an ad hoc list. Rather, there is a theory, codiﬁed in the periodic table of elements, that tells us how to enumerate these chemical kinds and how they are systematically related to each other. To say what the chemical kinds are, we can just consult this theory; we do not, in addition, have to go out and do ﬁeld work. No such theory exists in biology for species and higher taxa; ﬁeld work is basically the only method that science has for assembling a list of species. The terms ‘botanizing’ and ‘beetle collecting’ both allude to this feature of systematic biology. Species and higher taxa are things that happen to come into existence owing to the vagaries of what transpires in the branching tree of life. 335 ELLIOTT SOBER It does not follow that there are no natural kinds in evolutionary biology. Perhaps sexual reproduction is a kind; perhaps being a predator is another. What makes it true that two organisms each reproduce sexually, or that both are predators, is that they are similar in some respect; it is not required that they be historically connected to each other. The sexual species do not form a monophyletic group, and neither do the preda- tors. These kind terms appear in models of different evolutionary processes; there are models that explain why sex might evolve, and models that describe the dynamics of	Blackwell
predator–prey interactions. Although Darwin’s theory of evolution undermines essen- tialist interpretations of species and higher taxa, it is another matter whether essen- tialism is the right way to understand these other, non-taxonomic, theoretical concepts. If biological species lack essences, then Homo sapiens lacks an essence. If the essen- tialist is right about chemical kinds, then it makes sense to ask what the nature of gold is. But what could it mean to talk about ‘human nature’? Taken literally, the idea of human nature is the idea that there is an essence that all human beings, and only they, possess. This thought is at the root of ARISTOTLE’s (chapter 23) ethical theory. It also is at the root of many normative claims to the effect that this or that human behaviour is ‘unnatural’. This peculiar expression can be given a sensible biological reading, if ‘natural’ is simply equated with the idea of being found in nature. Understood in this way, everything that human beings do, both the good and the bad, is natural. However, when ‘natural’ and ‘unnatural’ are not used in this way, it is important to demand that an explanation be given of what the terms are supposed to mean. Normative ethical claims should not be permitted to masquerade as biology. 3.3 Evolutionary epistemology Evolutionary epistemology is a ﬁeld divided into two research programmes (Bradie 1994). First, there are attempts to use ideas from evolutionary biology to explain various features of human cognition. Second, there is the attempt to develop theories of cultural and scientiﬁc change that are analogous to theories of biological evolution. These lines of inquiry are independent of each other, so that the success or failure of the one does not automatically entail the success or failure of the other. Nor is either research programme particularly uniﬁed; this means, for example, that one evolution- ary model of scientiﬁc change might fail miserably, while another succeeds admirably. Sociobiology (Wilson 1975) and evolutionary psychology (see the papers collected in Barkow, Cosmides and Tooby 1992) provide examples of the ﬁrst type of evolutionary epistemology. Sociobiologists usually focus on different features of human behaviour and seek to explain them as adaptive responses to ancestral environments. The debate about adaptationism, discussed earlier, grew out of the vociferous controversy that engulfed sociobiology. Critics of sociobiology believed that sociobiology’s methodologi- cal defects were just the tip of the iceberg; the real problem, they thought, was the pervasive inﬂuence of naive adaptationism in evolutionary biology as a whole (Gould and Lewontin 1979). Evolutionary psychology differs from sociobiology, not by being less adaptationist, but by shifting its focus from behaviour to cognitive mechanisms as the items requiring evolutionary explanation. Evolutionary psychologists tend to think of the mind as a highly modularized set of devices, each having evolved as a solution to a different adaptive problem; they also tend to think that cognitive adaptations vary very 336 PHILOSOPHY OF BIOLOGY little among human beings. Both the modularity hypothesis and the universality hypothesis have been questioned (Sterelny and Grifﬁths 1999; Wilson 1994). This ﬁrst type of evolutionary epistemology has also been pursued by philosophers. For example, there has been considerable debate about whether, or in what circum- stances, natural selection can be expected to favour belief acquisition devices that are reliable – that is, that produce mostly true beliefs (Stephens 2001). There also has been investigation of when natural selection will favour innate and inﬂexible characteristics, and when it will favour traits that are learned and plastic (Godfrey-Smith 1991, 1996; Sober 1994b). Godfrey-Smith (1996) also explores the hypothesis that mentality evolved as a response to environmental complexity (Sober 1997b).	Blackwell
Examples of the second kind of evolutionary epistemology may be found in pro- posals developed by Popper (1973), Campbell (1974) and Hull (1988), which view scientiﬁc change as the result of a competition among theories, with the ﬁttest theory surviving. This does not mean that theories survive because they enable those who believe them to have more babies; nor do evolutionary epistemologists propose that scientiﬁc theories are passed from individual to individual by genetic transmission. Whereas models of biological evolution standardly measure ﬁtness in terms of repro- ductive output and treat genes as the mechanism of inheritance, evolutionary models of scientiﬁc change replace genetic transmission with teaching and learning, and think of ﬁtness as an idea’s propensity to spread. Fit ideas are attractive, regardless of whether they affect biological survival and reproduction (Sober 1993). There are some obvious disanalogies between biological models of evolution by natural selection and selectionist models of scientiﬁc change. For one thing, novel biological variation arises by random mutation; this means that what causes a mutation to occur has nothing to do with whether the mutation will be useful. In contrast, it seems clear that novel scientiﬁc ideas are often invented with the goal of being useful; scientists don’t make up hypotheses at random. However, this and other disanalogies do not undermine the claim that the ideas in a scientiﬁc community compete with each other; and that there is a selection process in which some ideas spread while others disappear from the scene. Thus, it seems clear that scientiﬁc change can be described as a selection process. The substantive task is to say how the concept of ﬁtness should be understood in this context. What makes one idea more attractive than another in a scientiﬁc community? Surely there are many factors that can inﬂuence an idea’s attractiveness, and the mix of features that make an idea attractive in one scientiﬁc context may differ from the mix that matters in another. Perhaps some episodes of scientiﬁc change are driven largely by observational evidence, while others occur because of religious, metaphysical or political commitments. Evolutionary epistemologists who want to model scientiﬁc change as a selection process must address the questions that historians of science debate when they consider ‘internal’ versus ‘external’ factors. Merely saying that scientiﬁc change occurs by a selection process is not enough. Another variety of evolutionary epistemology may be found in models of cultural group selection (Boyd and Richerson 1985). As Darwin observed in his discussion of human morality, groups compete with each other, and groups with more adaptive ideologies will out-compete groups whose ideas are less ﬁt. Again, there is no require- ment that these ideational elements are transmitted genetically; groups may vary simply because the individuals in one generation faithfully teach their mores to the next. Group 337 ELLIOTT SOBER competition may mean that people in one group kill the members of others, or it may simply mean that the conquerors impose their mores on the conquered. Indeed, the spread of ideas may not involve ‘conquest’ at all, if people in one group freely adopt ideas from another. 3.4 Evolutionary ethics HUME’s (chapter 31) distinction between is and ought – between claims that are purely descriptive and claims that are normative – must be borne in mind when we ask what relevance evolutionary theory has for questions about morality. Darwin suggested that evolutionary theory helps us understand why human morality has some of the features that it has. It remains to be seen how much of the content of human morality can be explained in this way, and how much should be thought of as arising by non-adaptive cultural processes. For example, some cultures impose restrictive norms on the cloth-	Blackwell
ing that men and women may wear, while others encourage personal innovation. Even if evolutionary theory has nothing to say about why this pattern of variation exists, it is possible that the theory has an important bearing on other patterns (Sober and Wilson 1998). What can evolutionary theory tell us about which normative ethical claims, if any, are true? Although many social Darwinists thought that evolutionary outcomes are always good (because they thought that evolution is inherently ‘progressive’), there has also been a strong tradition that maintains that evolutionary outcomes are often morally deplorable. Just as Darwin did not rejoice in the behaviour of parasitic wasps, so Huxley (1997) thought that the point of morality was to combat the instincts we inherit; more recently, and in the same vein, G. C. Williams (1989) has suggested that nature is a ‘wicked old witch’. Another line of thinking in evolutionary ethics holds that evolutionary considera- tions show that there can be no objective normative truths. The claim is not just that the moral principles we now happen to hold are mistaken; the claim is stronger – that no moral convictions can be true (Ruse and Wilson 1986). Evolution has created in us the illusion that there are objective moral standards; there is an adaptive advantage in believing this ﬁction. In reply, it should be noted that the mere fact that our moral beliefs are produced by evolution does not show that there are no moral truths. After all, our simple mathematical beliefs may be the result of evolution, but that does not show that there are no mathematical truths (Kitcher 1994). However, more needs to be said here, since the subjectivist does not have to maintain that the non-existence of moral truths deductively follows from the claim that the morality we accept is the product of evolution. A better formulation of subjectivism maintains that it is exceedingly improbable that the morality we accept is true, given that it is produced by evolution. After all, natural selection causes traits to evolve because they provide reproductive beneﬁts. Why should a process bent on maximizing reproduction lead us to moral beliefs that are true? The thing to notice about this line of thought is that it does not entail that there are no moral truths; rather, it advances the more modest hypothesis that the moral beliefs we presently have are not true. In addition, this argument fails to attend to the fact that evolution has given human beings the capacity to reason; if this capacity can lead us 338 PHILOSOPHY OF BIOLOGY to discover theoretical truths in science that provide no practical beneﬁt in terms of survival and reproduction, why should reason not also be able to lead us to discoveries about the good and the right? Perhaps there are reasons to doubt this, but the mere fact that natural selection aims to maximize reproduction should not lead us to embrace ethical subjectivism. A slightly different argument for ethical subjectivism maintains that we can provide a fully satisfactory explanation of why we behave the way we do, and also explain why we accept the moral principles we do, without postulating a realm of ethical truths (Harman 1977; Ruse and Wilson 1986). Objectivism about ethics is unparsimonious; a simpler explanation of human thought and behaviour traces these phenotypic features back to the social context of human culture and the biological context of human evolution. The right reply to this argument for subjectivism is that normative ethical principles are not in the business of explaining why people think and act as they do. The point of morality is to regulate behaviour, not to explain it. One might just as well argue that objective epistemological norms do not exist because they are not needed to explain why people think as they do. Psychology is in the business of explaining how people in fact think; epistemology, however, is in a normative line of work (Sober 1993).	Blackwell
Acknowledgements I am grateful to David Hull and Christopher Stephens for their help in the preparation of this chapter. References Allen, C., Bekoff, M. and Lauder, G. (eds) 1998: Nature’s Purposes: Analyses of Function and Design in Biology. Cambridge, MA: MIT Press. Anscombe, E. 1975: Causality and Determination. In E. Sosa (ed.) Causation and Conditionals. Oxford: Oxford University Press. Barkow, J., Cosmides, L. and Tooby, J. (eds) 1992: The Adapted Mind: Evolutionary Psychology and the Generation of Culture. Oxford: Oxford University Press. Beatty, L. 1995: The Evolutionary Contingency Thesis. In G. Wolters and J. Lennox (eds) Concepts, Theories, and Rationality in the Biological Sciences: The Second Pittsburgh–Konstanz Colloquium in the Philosophy of Science. Pittsburgh: University of Pittsburgh Press. Boorse, C. 1976: Wright on Functions. Philosophical Review, 85, 70–86. Boyd, R. and Richerson, P. 1985: Culture and the Evolutionary Process. Chicago: University of Chicago Press. Bradie, M. 1994: Epistemology from an Evolutionary Point of View. In E. Sober (ed.) Conceptual Issues in Evolutionary Biology, 2nd edn. Cambridge, MA: MIT Press. Brandon, R. 1978: Adaptation and Evolutionary Theory. Studies in the History and Philosophy of Science, 9, 181–206. —— 1990: Adaptation and Environment. Princeton, NJ: Princeton University Press. —— 1996: Concepts and Methods in Evolutionary Biology. Cambridge: Cambridge University Press. Buller, D. (ed.) 1999: Function, Selection and Design. Series in Philosophy and Biology. New York: State University of New York Press. Campbell, D. 1974: Evolutionary Epistemology. In P. Schilpp (ed.) The Philosophy of Karl Popper. La Salle, IN: Open Court. 339 ELLIOTT SOBER Cummins, R. 1975: Functional Analysis. Journal of Philosophy, 72, 741–64. Reprinted in E. Sober (ed.) (1994) Conceptual Issues in Evolutionary Biology, 2nd edn. Cambridge, MA: MIT Press. Darwin, C. 1981 [1871]: The Descent of Man, and Selection in Relation to Sex. Princeton, NJ: Princeton University Press. Dawkins, R. 1976: The Selﬁsh Gene, 2nd edn. Oxford: Oxford University Press. Desmond, A. and Moore, J. 1991: Darwin – the Life of a Tormented Evolutionist. New York: Times Warner. Dobzhansky, T. 1973: Nothing in Biology Makes Sense Except in the Light of Evolution. American Biology Teacher, 35, 125–9. Dupré, J. 1993. The Disorder of Things. Cambridge, MA: Harvard University Press. Edwards, A. and Cavalli-Sforza, L. 1964: Reconstruction of Evolutionary Trees. In V. Heywood and J. McNeill (eds) Phenetic and Phylogenetic Classiﬁcation. New York: Systematics Association Publication, 6, 67–76. Eldredge, N. and Cracraft, J. 1980: Phylogenetic Patterns and the Evolutionary Process. New York: Columbia University Press. Enç, B. 1986: Essentialism Without Individual Essences: Causation, Kinds, Supervenience and Restricted Identities. Midwest Studies in Philosophy, 11, 403–26. Ereshefsky, M. 1991: Species, Higher Taxa, and the Units of Evolution. Philosophy of Science, 58, 84–101. —— (ed.) 1992: The Units of Evolution: Essays on the Nature of Species. Cambridge, MA: MIT Press. Fisher, R. 1957 [1930]: The Genetical Theory of Natural Selection, 2nd edn. New York: Dover Books. Fox Keller, E. and Lloyd, E. (eds) 1992: Keywords in Evolutionary Biology. Cambridge, MA: Harvard University Press. Ghiselin, M. 1974: A Radical Solution to the Species Problem. Systematic Zoology, 23, 536–44. Godfrey-Smith, P. 1991: Signal, Decision, Action. Journal of Philosophy, 88 (12), 709–22. —— 1993: Functions: Consensus Without Unity. Paciﬁc Philosophical Quarterly, 74, 196–208. —— 1994: A Modern History Theory of Function. Nous, 28 (3), 344–62. —— 1996: Complexity and the Function of Mind in Nature. Cambridge: Cambridge University Press. Gould, S. 1980: The Panda’s Thumb. Harmondsworth: Penguin Books. —— 1989: Wonderful Life. New York: W. W. Norton. Gould, S. and Lewontin, R. 1979: The Spandrels of San Marco and the Panglossian Paradigm: A	Blackwell
Critique of the Adaptationist Programme. Proceedings of the Royal Society of London B, 205, 581–98. Reprinted in E. Sober (ed.) (1994) Conceptual Issues in Evolutionary Biology, 2nd edn. Cambridge, MA: MIT Press. Hamilton, W. 1967: Extraordinary Sex Ratios. Science, 156, 477–88. Harman, G. 1977: The Nature of Morality: An Introduction to Ethics. Oxford: Oxford University Press. Hennig, W. 1966: Phylogenetic Systematics. Urbana: University of Illinois Press. (Revision and translation of Hennig’s 1950 Grundzuge einer Theorie der phylogenetishen Systematik.) Hull, D. 1965: The Effect of Essentialism on Taxonomy: 2000 Years of Stasis. British Journal for the Philosophy of Science, 15, 314–26; 16, 1–18. —— 1974: Philosophy of Biological Sciences. Englewood Cliffs, NJ: Prentice-Hall. —— 1978: A Matter of Individuality. Philosophy of Science, 45, 335–60. Reprinted in E. Sober (ed.) (1994) Conceptual Issues in Evolutionary Biology, 2nd edn. Cambridge, MA: MIT Press. —— 1988: Science as a Process. Chicago: University of Chicago Press. Hull, D. and Ruse, M. (eds) 1998: Philosophy of Biology. Oxford: Oxford University Press. Hume, D. 1947 [1779]: Dialogues Concerning Natural Religion. New York: Thomas Nelson and Sons. 340 PHILOSOPHY OF BIOLOGY Huxley, T. 1997 [1893]: Evolution and Ethics. In A. Barr (ed.) The Major Prose of Thomas Henry Huxley. Athens, GA: University of Georgia Press. Kitcher, P. 1984: 1953 and All That: A Tale of Two Sciences. Philosophical Review, 93, 335–73. Reprinted in E. Sober (ed.) (1994) Conceptual Issues in Evolutionary Biology, 2nd edn. Cambridge, MA: MIT Press. —— 1993: The Advancement of Science. New York: Oxford University Press. —— 1994: Four Ways of Biologicizing Ethics. In E. Sober (ed.) Conceptual Issues in Evolutionary Biology, 2nd edn. Cambridge, MA: MIT Press. Kripke, S. 1972: Naming and Necessity. In G. Harman and D. Davidson (eds) The Semantics of Natural Language. Dordrecht: Reidel. Lakatos, I. 1978: Falsiﬁcation and the Methodology of Scientiﬁc Research Programmes. In The Methodology of Scientiﬁc Research Programmes: Philosophical Papers, vol. 1. Cambridge: Cambridge University Press. Lewontin, R. 1970: The Units of Selection. Annual Review of Ecology and Systematics, 1, 1–14. Lloyd, E. 1988: The Structure and Conﬁrmation of Evolutionary Theory. Westport, CT: Greenwood Press. Maynard-Smith, J. 1988: Evolutionary Progress. In M. Nitecki (ed.) Evolutionary Progress. Chicago: University of Chicago Press. Mayr, E. 1963: Animal Species and Evolution. Cambridge, MA: Harvard University Press. —— 1970: Populations, Species, and Evolution. Cambridge, MA: Harvard University Press. —— 1975: Typological versus Populational Thinking. In Evolution and the Diversity of Life. Cambridge, MA: Harvard University Press. Reprinted in E. Sober (ed.) (1994) Conceptual Issues in Evolutionary Biology, 2nd edn. Cambridge, MA: MIT Press. Mill, J. S. 1874: A System of Logic, 8th edn. Harper and Brothers. Mills, S. and Beatty, J. 1979: The Propensity Interpretation of Fitness. Philosophy of Science, 46, 263–88. Reprinted in E. Sober (ed.) (1994) Conceptual Issues in Evolutionary Biology, 2nd edn. Cambridge, MA: MIT Press. Mitchell, W. and Valone, T. 1990: The Optimization Research Program: Studying Adaptations by their Function. Quarterly Review of Biology, 65, 43–52. Nagel, E. 1961: The Structure of Science. London: Routledge and Kegan Paul. Oppenheim, P. and Putnam, H. 1958: Unity of Science as a Working Hypothesis. In H. Feigl, G. Maxwell and M. Scriven (eds) Minnesota Studies in the Philosophy of Science, vol. 2. Minneapolis: University of Minnesota Press. Orzack, S. and Sober, E. 1994: Optimality Models and the Long-run Test of Adaptationism. American Naturalist, 143, 361–80. —— 2001: Adaptationism, Phylogenetic Inertia, and the Method of Controlled Comparisons. In Adaptationism and Optimality. Cambridge: Cambridge University Press.	Blackwell
Paley, W. 1805: Natural Theology. Rivington. Popper, K. 1973: Objective Knowledge. Oxford: Clarendon Press. Putnam, H. 1975: The Meaning of Meaning. In K. Gunderson (ed.) Language, Mind and Knowledge: Minnesota Studies in the Philosophy of Science, vol. 7. Minneapolis: University of Minnesota Press. Rosenberg, A. 1978: The Supervenience of Biological Concepts. Philosophy of Science, 45, 368–86. —— 1985: The Structure of Biological Science. Cambridge: Cambridge University Press. —— 1994: Instrumental Biology, or the Disunity of Science. Chicago: University of Chicago Press. Royall, R. 1997: Statistical Evidence – A Likelihood Paradigm. London: Chapman & Hall. Ruse, M. 1973: The Philosophy of Biology. London: Hutchinson University Library. Ruse, M. and Wilson, E. 1986: Moral Philosophy as Applied Science. Philosophy, 61, 173–92. 341 ELLIOTT SOBER Reprinted in E. Sober (ed.) (1994) Conceptual Issues in Evolutionary Biology, 2nd edn. Cambridge, MA: MIT Press. Schaffner, 1976: Reductionism in Biology: Prospects and Problems. In R. S. Cohen et al. (eds) PSA 1974. Dordrecht: Reidel. Sober, E 1980: Evolution, Population Thinking, and Essentialism. Philosophy of Science, 47, 350–83. Reprinted in E. Sober (ed.) (1994) Conceptual Issues in Evolutionary Biology, 2nd edn. Cambridge, MA: MIT Press. —— 1984: The Nature of Selection. Cambridge, MA: MIT Press. (Second edition (1994) Chicago: University of Chicago Press.) —— 1988: Reconstructing the Past: Parsimony, Evolution, and Inference. Cambridge, MA: MIT Press. —— 1990: Putting the Function Back Into Functionalism. In W. Lycan (ed.) Mind and Cognition: A Reader. Oxford: Blackwell. —— 1992: Monophyly. In E. Fox Keller and E. Lloyd (eds) Keywords in Evolutionary Biology. Cambridge, MA: Harvard University Press. —— 1993: Philosophy of Biology. Boulder, CO: Westview Press. (Second edition 1999.) —— 1994a: Progress and Direction in Evolution. In J. Campbell and W. Schopf (eds) Creative Evolution?! Boston: Jones and Bartlett. —— 1994b: The Adaptive Advantage of Learning and a priori Prejudice. In From a Biological Point of View. Cambridge, MA: Cambridge University Press. —— 1997a: Two Outbreaks of Lawlessness in Recent Philosophy of Biology. In L. Darden (ed.) PSA 1996, 458–67. —— 1997b: Is the Mind an Adaptation for Coping with Environmental Complexity? Biology and Philosophy, 12, 539–50. —— 1999a: Physicalism from a Probabilistic Point of View. Philosophical Studies, 95 (1–2), 135–74. —— 1999b: The Multiple Realizability Argument Against Reductionism. Philosophy of Science, 66 (4). —— 1999c: Modus Darwin. Biology and Philosophy, 14 (2), 253–78. —— 1999d: Testability. Proceedings and Addresses of the American Philosophical Association, 73 (2), 47–76. Sober, E. and Lewontin, R. 1982: Artifact, Cause, and Genic Selection. Philosophy of Science, 47, 157–80. Sober, E. and Wilson, D. 1994: A Critical Review of Philosophical Work on the Units of Selection Problem. Philosophy of Science, 61, 534–55. Reprinted in D. Hull and M. Ruse (eds) Philosophy of Biology. Oxford: Oxford University Press. —— 1998: Unto Others – the Evolution and Psychology of Unselﬁsh Behaviour. Cambridge, MA: Harvard University Press. Stephens, C. 2001: When is it Selectively Advantageous to Have True Beliefs? Sandwiching the Better Safe than Sorry Argument. Philosophical Studies, 105, 161–89. Sterelny, K. and Grifﬁths, P. 1999: Sex and Death: An Introduction to the Philosophy of Biology. Chicago: University of Chicago Press. Sterelny, K. and Kitcher, P. 1988: The Return of the Gene. Journal of Philosophy, 85, 339–61. Wade, M. 1978: A Critical Review of Models of Group Selection. Quarterly Review of Biology, 53, 101–14. Waters, C. K. 1990: Why the Anti-reductionist Consensus Won’t Survive the Case of Classical Mendelian Genetics. PSA 1990, vol. 1, 125–39. Reprinted in E. Sober (ed.) (1994) Conceptual Issues in Evolutionary Biology, 2nd edn. Cambridge, MA: MIT Press.	Blackwell
Wiley, E. 1981: Phylogenetics: The Theory and Practice of Phylogenetic Systematics. London: John Wiley. 342 PHILOSOPHY OF BIOLOGY Williams, G. 1966: Adaptation and Natural Selection. Princeton, NJ: Princeton University Press. —— 1989: A Sociobiological Expansion of Evolution and Ethics. In J. Paradis and G. C. Williams (eds) T. H. Huxley Evolution and Ethics: With New Essays on its Victorian and Sociobiological Context. Princeton, NJ: Princeton University Press. —— 1992: Natural Selection: Domains, Levels and Challenges. Oxford: Oxford University Press. Wilson, D. 1980: The Natural Selection of Populations and Communities. Menlo Park, CA: Benjamin Cummings. —— 1994: Adaptive Genetic Variation and Human Evolutionary Psychology. Ethology and Sociobiology, 15, 219–35. Wilson, E. 1975: Sociobiology: The New Synthesis. Cambridge, MA: Harvard University Press. Wimsatt, W. 1979: Reduction and Reductionism. In P. Asquith and H. Kyburg (eds) Current Research in Philosophy of Science. Philosophy of Science Association, 352–77. —— 1980: Reductionistic Research Strategies and their Biases in the Units of Selection Controversy. In T. Nickles (ed.) Scientiﬁc Discovery, vol. 2. Dordrecht: Reidel. Wright, L. 1973: Functions. Philosophical Review, 82, 139–68. Reprinted in E. Sober (ed.) (1994) Conceptual Issues in Evolutionary Biology, 2nd edn. Cambridge, MA: MIT Press. —— 1976: Teleological Explanations: An Etiological Analysis of Goals and Functions. Berkeley: University of California Press. Discussion Questions 1 What constitutes being alive? 2 How should we understand ﬁtness? 3 If biological properties supervene upon physical properties, can biology be reduced to physics? 4 Is reductionism compatible with the multiple physical realizability of biological properties? 5 What is required for one theory to be reducible to another? 6 How should we understand the role of function in biology? 7 Must we take account of our explanatory interests to distinguish between the function and the fortuitous beneﬁt of an organ? 8 Are both Newton’s theory of gravitation and Darwin’s theory of evolution scientiﬁc? 9 How can we determine which species are closely related and which are related only more distantly? 10 What is the role of parsimony in scientiﬁc reasoning? 11 Can we explain why natural selection does not always result in species having optimal traits? 12 Is natural selection merely one among several important inﬂuences on trait evolution, or is it the only important inﬂuence? 13 What is the difference between testing hypotheses and testing research programmes? 14 Why in biology is it important to recognize that historical origin and present advantage are not automatically linked? 15 Do more than one unit of selection have a role in the evolutionary process? 16 Is the choice between individuals and groups as the units of selection a matter of adopting a convention rather than a matter of answering a factual question about the history of life? 343 ELLIOTT SOBER 17 Are there any laws of evolution? 18 How can a priori models have a part to play in empirical science? 19 Does the adaptive complexity of organisms justify the argument from design? 20 Could species, as understood in evolutionary theory, have essences? 21 What makes human beings now and human beings thousands of years ago members of the same species? 22 Are species individuals or natural kinds? What could count as natural kinds in evolutionary biology? 23 What follows from the claim that species are historical entities? 24 Is evolutionary psychology more securely based than sociobiology? 25 Is biological evolution a good model for explaining social and cultural change? 26 What are the consequences for philosophy of accepting a modularity account of the mind? 27 Can natural selection help to explain the existence and character of human mentality? 28 Does an evolutionary theory of the character of human morality help us to assess	Blackwell
moral claims? 29 Do evolutionary considerations undermine the claim that there are objective moral truths? 344 11 Philosophy of Mathematics M A RY T I L E S Since the time of ancient Greece, mathematics has been intimately tied to philosophy, both as a model of knowledge and as an object of philosophical reﬂection. Are numbers real? What is a proof? Is mathematics more certain than other knowledge? Can ﬁnite minds have knowledge of inﬁnity? How can mathematics apply to the world? In this chapter, changing philosophical conceptions of mathematics, changes in the historical context of mathematical thought and changes in mathematics itself are explored in relation to a basic question: how can theoretical reasoning about non-concrete mathe- matical objects be both secure and so useful? Platonic, Aristotelian and Kantian approaches to mathematics are examined in their own right and to place in context the problems and proposed solutions of the major modern schools of logicism, formalism and intuitionism. Recent developments which do not seek foundations for mathematics are also considered. Many of the discussions of great historical ﬁgures in this volume, especially FREGE AND RUSSELL (chapter 37), are relevant to the present chapter. Readers will also wish to consult chapters on EPISTEMOLOGY (chapter 1), METAPHYSICS (chapter 2), PHILOSOPHY OF LANGUAGE (chapter 3), PHILOSOPHY OF LOGIC (chapter 4) and PHILOSOPHY OF SCIENCE (chapter 9). Introductions to the philosophy of mathematics often begin where Körner’s (1960) inﬂuential introduction began, outlining three positions: logicism, formalism and in- tuitionism. These were the three contending schools to emerge from nineteenth-century mathematical moves to provide rigorous foundations for mathematical analysis (including inﬁnitesimal calculus). The problem for the philosophy of mathematics has been that (1) these seemed to represent all the reasonable positions available, and (2) in the light of Gödel’s incompleteness theorems and other results proved by Turing, Church, Skolem and Tarski in the 1930s, neither logicism nor formalism seemed a philosophically viable position. Intuitionism, whilst having its philosophical credentials intact, was unacceptable to most mathematicians because it involved discarding parts of classically accepted mathematics. This apparent impasse partly explains the decline of interest in the philosophy of mathematics since the ﬁrst part of this century, when for a while, with the work of Russell and Whitehead, and the Vienna Circle logical positivists, it seemed to occupy centre stage. MARY TILES Hao Wang, in his perceptive retrospective analysis of the philosophy of mathe- matics of this period (Wang 1968), explains this trajectory in terms of the wider move- ments of analytic philosophy, which were initially strongly empiricist and which consequently inherited a hostility toward abstract objects. Mathematics presents a signiﬁcant challenge to empiricism, as was made painfully evident by the prominence of mathematics in Einstein’s relativity theories and in the development of quantum mechanics. How is empiricism to give an adequate explanation of the certainty, clarity, universality and applicability of mathematics? Wang’s answer is that the above mentioned developments in logic and the foundations of mathematics show that analytic empiricism cannot meet this challenge. It is therefore only as philosophy in general begins to move beyond analytic empiricism that more fruitful approaches to philosophy of mathematics can begin to be explored. But how exactly is such a move to be made? Because the three foundational schools originated in the work of philosophically minded mathematicians who were seeking to resolve sophisticated conceptual problems that had arisen in the context of ongoing mathematical research, much of the work in foundations is itself technical and mathe-	Blackwell
matical in character. There is no sharp line where mathematics ends and philosophy begins. Indeed, one of the striking features of the development of mathematics in the twentieth century was the way in which attempts to solve philosophical and con- ceptual problems generated new mathematics. Thus from the standpoint of the early twenty-ﬁrst century, work in the foundations of mathematics can by no means be ignored, but it may need to be re-evaluated. What is its status? What exactly does it tell us about contemporary mathematics? These questions must now be added to the philo- sophical agenda. So one of the ways of moving beyond the presuppositions built into philosophy of mathematics by analytic empiricists is to recontextualize their work on logic and foundations in the hope of understanding its motivations, achievements and shortcomings. A ﬁrst step is to sketch, using broad strokes, major features in the land- scape of Western philosophical reﬂection on mathematics, many of which were already put in place by the ancient Greeks. 1 Basic Tasks for a Philosophy of Mathematics The basic puzzle, of which the challenge to empiricism is but a specialized variant, is to explain how theoretical reasoning about highly abstract objects can be so useful. The features of mathematics which seemed to mark it off from other disciplines are that its objects are not encountered in the world of sense experience, and yet, perhaps as a con- sequence, mathematical claims can be provided with proofs which seem to establish them with an exactness and certainty unparalleled in other branches of knowledge. But how is it that people shut away from the outside world, thinking, calculating and proving theorems about things which we can never hope to see or touch (numbers such as 2, 1 million, p, or structures such as circles, rings and vector spaces), produce results which other, very practical people ﬁnd so useful in their dealings both with the every- day world and with the esoteric worlds of high-tech science? What right have we to be so conﬁdent of the correctness of these results? 346 PHILOSOPHY OF MATHEMATICS A x O B O x C E D Figure 11.1 Proof that the internal angles of a triangle add up to the sum of two right angles. The straightforward answer to this last question is that mathematicians are able to offer proofs of their results. For example, as geometers before the time of Aristotle real- ized, measurements of the internal angles of plane triangles do not yield one exact result, but results that cluster round two right angles, and, because measuring gives no insight into why this clustering should occur, it affords no assurance that a ‘rogue’ triangle might not some day be found whose internal angles had a markedly different sum. A proof, on the other hand, that shows why, given what it is to be a triangle, its internal angles must add up to two right angles, assures us that no rogue triangle exists. One such proof is shown in ﬁgure 11.1. Such a proof is possible only if it can be assumed that all the lines involved are per- fectly straight and have no thickness (something which is not true of the lines in the diagram). Lines which have thickness do not meet in points and if they have wriggles in them there is no sense to be given to an exact measurement of the angle between them. So even to have internal angles whose sum is exactly equal to two right angles is possible only for a perfect plane triangle, which we could not encounter in experi- ence. The theorem proved below will hold more or less exactly for those triangles we encounter to the extent that they are more or less perfectly triangular. There is an important assumption made here about the applicability of results proved about idealizations: a result proved about an idealization will hold for situations which approximate to that ideal in direct proportion to the closeness with which they approximate to the ideal. This assumption has traditionally underwritten applications	Blackwell
of mathematics, but, with recent work in chaos theory, it has been shown not to be valid for systems exhibiting sensitive dependence on initial conditions (Stewart 1989: chs 7, 14). This work opens up new questions about the connection between proofs and the use of results proved. But how can we tell that a purported proof really is a proof? All proofs have to start from some assumptions (we assumed above some properties of parallel lines and of addition of angles). How can we be certain that these are correct? Furthermore, how can we be sure that all inferences are made correctly? To ask for a proof for every assumption will lead to an inﬁnite regress. So if there are to be any proofs there must be some self-evident starting-points, which can be known to be necessarily true and which are neither capable of, nor stand in need of, further justiﬁcation. 347 MARY TILES This means that if mathematics is a discipline in which proofs are required then it must be one which exhibits a systematic hierarchical order. Some propositions con- cerning mathematical entities will appear as ﬁrst principles, accepted without proof: these are the foundations upon which subsequent knowledge must be built. Basic the- orems are proved from ﬁrst principles: these theorems are in turn used to prove further results, and so on. This is the sense in which, so long as mathematicians demand and provide proofs, their discipline must necessarily be organized along lines approximat- ing the pattern already to be found in the Elements of Geometry of Euclid (which dates from around 300 BC) – a work which for centuries served as an exemplar. The question of what principles of inference are generally valid (rationally com- pelling, or truth-preserving) is traditionally assigned to LOGIC (chapter 4). But one may wonder whether there are not some speciﬁcally mathematical principles of inference. Take, for example, Euclid’s principle that if A = B and C = D, then A + C = B + D. Is this a logical principle (a speciﬁc application of the SUBSTITUTIVITY OF IDENTICALS (p. 800))? Is it a basic principle of the theory of part and whole? If so, does this belong to logic or to mathematics? Or is it a speciﬁcally mathematical principle governing extensive magnitudes? At this point it is already possible to see how concern for principles of valid deduc- tive inference and for the nature and status of ﬁrst principles becomes a natural part of any philosophy of mathematics that takes proof to be that on the basis of which mathematical propositions gain their aura of privileged certainty, or necessary truth. However, to equate the philosophy of mathematics with studies in logic and foundations would seem to ignore one half of the combination of features that make mathematics so philosophically puzzling and intriguing. It leaves out questions concerning the utility of mathematics. So far we have the following schema of questions structuring the problematic of philosophical reﬂections on mathematics: Basic Problem How can theoretical reasoning about non-concrete mathematical objects be both so secure and so useful? If the question of security is presumed to be answered by the fact that theoretical reasoning produces proofs of mathematical propositions, then we have to distinguish between ﬁrst principles and results proved from them, to get something like the following subsidiary problems: 1 What is the nature and status of the ﬁrst principles of mathematics? 1.1 Can its ﬁrst principles be identiﬁed? 1.2 How can these ﬁrst principles be known with certainty? 2 How is mathematical knowledge acquired? 2.1 Discovery: How are discoveries made? How are results established or proofs found? 2.2 Justiﬁcation: What constitutes a proof? What principles of inference are available? What is their nature and what grounds have we for thinking them reliable? 348 3 Given the answers under (1) and (2), how is the applicability of proved mathemat-	Blackwell