Notes

1
The Millian Philosophy

1 The verdict came even from those in greatest sympathy with him. Henry Sidgwick conceded that ‘Mill will have to be destroyed, as he is becoming as intolerable as Aristeides, but when he is destroyed, we shall have to build him a mausoleum as big as his present temple of fame’. Quoted in Winch (1970:47).

2 The most important figure was probably John Stirling. It seems likely that Mill had not actually read Kant when he wrote the System of Logic (see his letter to Comte, 13 March 1843; XIII 574). Letters of 1828 and 1829 (XVII 1954, 1956) to Thomas Wirgman, author of the Principles of the Kantesian or Transcendental Philosophy (London, 1824), give evidence of familiarity with Kant’s philosophy but no definite evidence of reading Kant himself. By the time of the Examination of Sir William Hamilton’s Philosophy Mill’s knowledge of Kant is much more detailed: he recognises ‘the whole difference of level which has been gained to philosophy through the powerful negative criticism of Kant’ (X 1), and refers frequently to positions taken by him.

3 Mill on Bentham’s narrowness of vision:

He was a boy to the last. Self-consciousness, that daemon of the men of genius of our time, from Wordsworth to Byron, from Goethe to Chateaubriand, and to which this age owes so much both of its cheerful and its mournful wisdom, never was awakened in him. How much of human nature slumbered in him he knew not, neither can we know. He had never been made alive to the unseen influences which were acting on himself, nor consequently on his fellow creatures. Other ages and other nations were a blank to him for purposes of instruction. He measured them but by one standard; their knowledge of facts, and their capability to take correct views of utility…. (‘Bentham’, X 92)

But a place was to be assigned to Bentham among the ‘masters of wisdom’; ‘he was not a great philosopher, but he was a great reformer in philosophy’ (X 82, 83). He epitomised the enlightenment’s negative critique: ‘he is the great subversive, or in the language of continental philosophers, the great critical, thinker of his age and country’ (X 79). Hart (1982: introduction and essay 1, The Demystification of the Law’) compares demystifying critique in Bentham and in Marx. Harrison shows how Bentham’s assault on mystifying and poisoned speech led into his theory of ‘fictions’.

4 Described in the Autobiography (I 137–53), and much over-interpreted ever since.

5 Cf. the Autobiography:

I never…wavered in the conviction that happiness is the test of all rules of conduct, and the end of life. But 1 now [after his mental crisis] thought that this end was only to be attained by not making it the direct end. Those only are happy (I thought) who have their minds fixed on some object other than their own happiness; on the happiness of others, on the improvement of mankind, even on some art or pursuit, followed not as a means, but as itself an ideal end. Aiming thus at something else, they find happiness by the way. The enjoyments of life…are sufficient to make it a pleasant thing, when they are taken en passant, without being made a principal object. Once make them so, and they are immediately felt to be insufficient. They will not bear a scrutinising examination. Ask yourself whether you are happy, and you cease to be so. (I 145–7)

Mill compares this to Carlyle’s ‘anti-self-consciousness theory’. It applies, he thinks, to all ‘those who have but a moderate degree of sensibility and of capacity for enjoyment, that is, for the great majority of mankind’.

6 This part of the Millian scheme has recently been very fully explored by a number of scholars. See for example the articles by Lyons (1976; 1978; 1982); and most fully, Berger (1984). Chapter 4 of Berger’s study gives an extended survey of Mill’s treatment of ‘natural impressions’ of justice— fairness, equality and, particularly in connnection with punishment and with economic justice, desert.

7 …the Germano-Coleridgean doctrine is…the revolt of the human mind against the eighteenth century, It is ontological, because that was experimental; conservative, because that was innovative; religious, because so much of that was infidel; concrete and historical, because that was abstract and metaphysical; poetical, because that was matter-of-fact and prosaic…. (‘Coleridge’ X 125)

8 The basic principles of associationism are outlined in 8.4.

9 There is a careful account of Bentham’s views on rights, fictions and paraphrasis in Harrison (1983:60ff. Cf. Hart 1982: ‘Natural Rights: Bentham and John Stuart Mill’). Bentham held rights to be ‘fictions’, but did not thereby mean to dismiss them:

from the observation, by which, for example, the words duties and rights are here spoken of as names of fictitious entities, let it not for a moment so much as be supposed, that, in either instance, the reality of the object is meant to be denied, in any sense in which in ordinary language the reality of it is assumed. (Quoted in Harrison 1983:99)

10 Cf. Scanlon (1984). Note that a philosophical utilitarian who acknowledges baseline constraints on maximisation of aggregate welfare does, in one sense, thereby acknowledge individual rights as primitive, But he is still not endorsing ‘natural rights’ in the sense in which they are dismissed by Bentham—as metaphysical grounds of obligation; he is simply imposing one among a number of possible distributive structures on the general good.

11 Clear statements of sceptical contractualism can be found, in political economy, in Buchanan (1975), and in recent philosophy, in Mackie (1977).

12 …utilitarianism (socialism, democracy) criticizes the origin of moral evaluations, but it believes them just as much as the Christian does.

(Naïveté: as if morality could survive when the God who sanctions it is missing! The ‘beyond’ is absolutely necessary if faith in morality is to be maintained.) (Nietzsche 1968: para. 253)

Nietzsche found this guilelessness, or alternatively, hypocrisy or self-deception, to be peculiarly English: ‘one still believes in good and evil and experiences the triumph of the good and the annihilation of evil as a task (that is English; typical case: the flathead John Stuart Mill)…. (para. 30). Compare also paras 925, 926 and, on the same theme, Daybreak (Nietzsche 1982), e.g. para. 132.

13 Mill judged it ‘the best chapter in the two volumes’ in a letter to R.B. Fox (14 February 1843, XIII 569). Compare also his letter to de Tocqueville of 3 November 1843 (XIII 612).

14 His opposition to the secret ballot is one index of this—the vote was not just an individual right but a public trust, to be exercised in public.

2
The analysis of language

1 The basic ideas were developed by Mill in his early twenties at meetings of the Society of Students of Mental Philosophy. This was a small discussion group which met twice a week in the mornings before work (8.30 to 10.00 a.m.), at George Grote’s house in Threadneedle Street. It discussed political economy and logic. See Mill’s Autobiography.

Books i and ii of the System, dealing with the analysis of language, and the epistemology of deductive inference, were the first to be written. The earliest drafts date from 1830 or 1831- A first draft of Book iii, ‘Of Induction’, was completed in 1837; drafts of the remaining three books were completed by the autumn of 1840. So the System is a work of Mill’s twenties and early thirties.

The evidence for these dates is collected in the Textual Introduction to the Collected Works edition of the System.

2 A good many of his remarks would fit the view that the objects of propositional attitudes are possible propositions—taking a proposition to be a token sentence. Mill often talks of ‘possible propositions’. On this view, the objects of the attitudes are modal constructs: permanent possibilities of utrerance or inscription. I seem to quantify over these constructs when, for example, 1 say, ‘Smith believes something which no one has ever written or said’. But on the ‘construct’ view, what I am saying is that a sentence could be constructed which would express Smith’s belief.

It would be natural to use the term ‘proposition’ to refer to the constructs; but that is not Mill’s use of the term.

3 All too often Mill uses an ordinary word in a technical or formal sense and then continues using it informally:

The predicate is the name denoting that which is affirmed or denied. The subject is the name denoting the person or thing which something is affirmed or denied of. The copula is the sign denoting that there is an affirmation or denial; and thereby enabling the hearer or reader to distinguish a proposition from any other kind of discourse.

In the italicised occurrences ‘denoting’ occurs informally. In Mill’s strict usage the predicate does not denote but connotes what is affirmed or denied, and the copula neither denotes nor connotes. Overall, Mill is thoroughly loose and inconsistent in his use of such words as ‘name’, ‘signify’, ‘denote’, ‘imply’ —sacrificing exactness to readability. I have silently tidied up.

4 Mill distinguished general from collective names:

A general name is one which can be predicated of each individual of a multitude; a collective name cannot be predicated of each singularly, but only of all taken together. The 76th Regiment of foot in the British Army; which is a collective name, is not a general but an individual name; for though it can be predicated of a multitude of individual soldiers taken jointly, it cannot be predicated of them severally. (VII 28)

5 The programme is fraught with difficulty. To mention one: consider ‘White is a colour’. Does that mean the same as ‘Whatever is white is coloured? It means more: it tells one that a white thing is, in virtue of its whiteness, coloured.

6 Compare Kripke’s distinction between ‘fixing a reference’ and ‘giving a meaning’, in Kripke (1980).

7 The problem exercised the associationists, for obvious reasons. Mill’s friend, Alexander Bain, was one of the first to treat it by analysing belief dispositionally. His analysis, through Peirce, had some influence on pragmatism. For Mill’s objections to it, see his critical notice of 1859, ‘Bain’s Psychology’ (XI 339–74).

8 Mill also discusses a third view, which takes general names to ‘denote’ classes of objects (with corresponding modifications to (c)). But he says that this differs from Nominalism only in being more obscure, and restricts his criticism to an attack on the unclarity of the notion of a class.

9 Compare Reginald Jackson (1941:34, footnote 2): The fundamental mistake is the failure to distinguish whether “s” from whether “s” is true. Whateley makes the same mistake (Elements of Logic, iv, iv, sect.1). Who does not?’ Jackson’s book contains many accurate and perceptive comments on Mill’s logic.

10 The rule for a connotative singular name—e.g. ‘the father of Socrates’ — would again specify the denotation conditionally, along the following lines: ‘For any object x, “the father of Socrates” denotes x if Socrates is a son of x and there is no object y, such that x is not identical with y and Socrates is a son of y.’ But in fact there could not be such a dictionary entry for every connotative singular name, since their number is infinite. One needs something like Russell’s theory of descriptions, functioning as a compositional rule. There is, it need hardly be added, nothing about this point in Mill.

11 I am indebted in this paragraph to McDowell (1977).

The ‘pure’ Millian line would presumably be that the two names have the same semantic content. The difficulty suggested in the text is that where proper names have the same semantic content they must be everywhere substitutable without loss of meaning. Consider the following two sentences:

Gorbachev believes that George Orwell was a novelist.

Gorbachev believes that Eric Blair was a novelist.

May not the sentences differ in truth-value? If they can, ‘George Orwell’ and ‘Eric Blair’ must differ in semantic content.

If we follow McDowell’s suggestion we would say that the names differ in semantic content, despite denoting the same thing, because semantic content is fixed by the denotation rule and their denotation rules differ. In contrast, the pure Millian line must say that the sentences do after all coincide in semantic content and truth-value (in so far as their content is clear at all). They could still, by pragmatic but easily intelligible conventions, and with some indeterminacy, be used to ascribe to Gorbachev non-overlapping ranges of beliefs. Cf. Kripke (1979). The issue cannot be pursued further here.

12 Not strictly true. Having noted that ‘a conditional proposition is a proposition concerning a proposition’ (VII 83), Mill adds that other attributes may be predicated of propositions as well, and gives examples:

We may say, That the whole is greater than its part, is an axiom in mathematics: That the Holy Ghost proceeds from the Father alone, is a tenet of the Greek Church: The doctrine of the divine right of kings was renounced by Parliament at the Revolution: The infallibility of the Pope has no countenance from Scripture. In all these cases the subject of the predication is an entire proposition. (VII 84)

A few pages earlier Mill comments on modal propositions:

Caesar may be dead; Caesar is perhaps dead; it is possible that Caesar is dead; … [are] properly asserted not of anything relating to the fact itself, but of the state of our own mind in regard to it; namely our absence of disbelief of it. Thus ‘Caesar may be dead’ means 1 am not sure that Caesar is alive’. (VII 81)

Mill’s view of what these sentences would ordinarily be taken to mean is plainly correct. They assert epistemic, not metaphysical modalities. As we shall see, Mill’s view is that the supposed ‘metaphysical’ distinction between necessity and possibility is empty of content.

13 Not that he was uninterested in language considered, so to speak, as an anthropological phenomenon. Far from it. Chapters iv.iv-vi, which treat of ‘the requisites of a philosophical language’ and of ‘the natural history of the variations in the meaning of terms’ are particularly interesting; e.g. the Coleridgean reflections on the ‘evil consequences of casting off any portion of the customary connotations of words’. Truths which have degenerated into dead dogmas can often, Mill thinks, be revived by the imaginative recovery of old meanings in the words which express them. But the process is imperilled by ‘the shallow conceptions and incautious proceedings of mere logicians’ who want to tidy up language (iv.iv.6). Mill’s most essential preoccupations—in this case the loss of old truths not by rejection but by over-ready acceptance—regularly turn up in unexpected places.

14 Michael Dummett distinguishes a sense and a reference version of the context principle. He presents the latter as a philosophical defence of mathematical platonism. The defence, as elaborated by Dummett (and in greater detail in Wright 1983) is deep and challenging. But I confess that I cannot find it in Frege. See Dummett (1981).

15 The Millian empiricist would have to propose a nominalistic treatment of set theory itself. A Millian way forward, in terms of a theory of the collectings and orderings of an idealised agent, is suggested by Kitcher (1983: e.g. 139).

3
Verbal propositions and apparent inference

1 Mill’s first complete draft of Book i (1831–2) of the System of Logic did not contain his theory of kinds. He was led to it, he says in his Autobiography (I 191, 229), only when in 1838 he completed the draft of Book iii, on induction:

In working out the logical theory of those laws of nature which are not laws of Causation, nor corollaries from such laws, I was led to recognise Kinds as realities in nature, and not mere distinctions for convenience; a light which I had not obtained when the first book was written, and which made it necessary for me to modify and enlarge several chapters of that book. (I 229)

Mill is referring to iii.xxii, ‘Of Uniformities of Coexistence not Dependent on Causation’. Properties may be found uniformly coexisting, he thinks (e.g. the various properties of water), without that fact being explicable by reference to a common cause. Objects which instantiate such clusters of properties constitute natural kinds. This is exactly opposite to the Lockean notion of a natural kind, where the multitude of perceptible properties uniformly instantiated by objects belonging to a natural kind is regarded as traceable to an underlying constitution or ‘real essence’ which explains their co-presence. Mill notes this Lockean conception of real essence as ‘corpuscular structure’ at i.vi.3 but does not connect it with the idea of a natural kind. (And he seems to agree with Locke’s conception in a footnote in the early editions of the System (VII 112).)

2 If our account in 2.8 of his view of negation is correct, then Mill ought to say that in a sentence like ‘No bachelors are married’ the attribute denied is already denied in the subject name.

3 Mill’s picture seems to involve Strawsonian distinctions between a sentence, the use of a sentence, and the assertion made by its use. It also seems to require a free logic for handling the semantics of the sentence itself. This does not fit with his analysis of the import of propositions, as we noted in 2.8 —there is often a conflict in Mill’s thinking betwen intuitions based on ordinary language and syllogistic principles, and constraints suggested by the attempt to think systematically about semantic concepts.

4 Particular propositions (‘Some vixens are foxes’), and singular propositions (The vixen in the barn is a fox’), can be handled along similar lines. In each case they will contain a verbal ingredient: ‘if x is a vixen then x is a fox’, and a non-verbal existential ingredient: some things are vixens, there is one and only one vixen in the barn….

5 See White (1978).

6 On Mill’s knowledge of Kant see note 2 to chapter 1.

7 It appears that Kant did not. He remarks that

Analytic judgements really teach us nothing more about the object than what the concept which we have of it already contains; they do not extend our knowledge beyond the concept of the object, but only clarify the concept. (Critique A736, B764)

This statement, when applied to judgements which are analytic in the narrow sense, sufficiently explains how such judgements can be known a priori—they are empty of cognitive content. But it does not explain how logical principles in general can be regarded as empty of cognitive content. That the criterion of what we have called ‘connotative inclusion’, or in Kant’s terms, of the ‘containment’ of one concept in another, could show this to be so is not obvious, and further analysis shows that it is not the case, So anyone who wants to continue to claim that logic is empty of cognitive content must supply some other way of showing it to be so.

Kant however took the principle of contradiction to be ‘without content and merely formal’ (Critique A152, B191). And he held that noumena could be thought, and that the laws of logic provided, as Charles Parsons says (Parsons 1969), a ‘negative criterion of truth’ in our statements about things in themselves. Had Kant endorsed Mill’s claim that logic as well as mathematics contains real propositions he would have had to take a transcendental idealist view of the aprioricity of logic just as he did of arithmetic and geometry, and hence would have had to conclude that it holds only of things ‘as they are cognisable by us’.

The followers of Kant whom Mill discusses in the Examination also believed that logic holds of ‘things in themselves’. A lot of Mill’s difficulty in getting his position clearly and consistently expressed comes from the fact that he wants to agree that deductive logic is in some sense formal, but at the same time to criticise the view that it can be assumed to hold of things in themselves. The criticism is just only if logic has empirical content, and so ‘formal’ cannot mean ‘empty of empirical content’.

8 The classic texts which have recently brought out the real strength of essentialism are by Kripke (1971; 1980). Space precludes any full comparison of Mill’s and Kripke’s views. According to Locke (i) all terms have information content, (ii) there are real essences. According to Mill, (i) no proper names have information content, (ii) there are no real essences. Thus Kripke agrees with Mill on the ‘semantic’ point and Locke on the ‘metaphysical’ point.

9 The meaning of a non-connotative abstract name is declared by spelling out the attributes it denotes.

10 Mill only considers conjunctively complex names, such as ‘vixen’. One can treat ‘vixen’ as conjunctively connoting each of the attributes, being female and being a fox. But a ‘negative name’, e.g. ‘immortal’, would have to be analysed as negatively connoting the attribute of mortality. A disjunctively complex name (examples are hard to find) would disjunctively connote its attributes. In short, in declaring the meaning of a connotative name, one must specify its syntactic contribution to determining the structure of propositions in which it occurs. One specifies, given Mill’s theory of negation, that ‘S is immortal’ would negatively predicate mortality of S. In the case of a disjunctively complex name, one specifies that predicating it constitutes the assertion of a compound disjunctive proposition. Even then Mill’s picture of semantic analysis remains radically incomplete, most notably because he has no proper theory of relational expressions.

The method of analysing the meaning of names by listing the attributes connoted is in any case open to objection. It makes the question, whether a connotative name is simple or complex, appear to turn on an issue of ontology—what attributes are there? (Is there an attribute of mortality? Or of immortality?) Ignoring the difficulties introduced by abstract names, this appearance can be avoided by following our alternative ‘Millian’ account in chapter 1. We then ‘declare the meaning’ of ‘vixen’ as follows: “‘vixen” denotes an object if and only if that object is female and is a fox’; and of ‘white’ as follows: “‘white” denotes an object if and only if that object is white’. Instead of saying that ‘white’ is semantically simple, because it connotes a single attribute, we say that it is simple because the denotation rule for ‘white’ uses the word in stating the condition which an object denoted by ‘white’ must meet. (Assuming that the denotation rules for the language are given in the same language.) Simplicity is relative to the resources of a language. And a name will be semantically complex if the proposition on the right-hand side in the denotation rule, which states the condition for an object’s satisfying the name, is compound.

11 For example as between the following two passages:

a class is absolutely nothing but an indefinite number of individuals denoted by a general name (VII 93)

when…we discover that…attributes are possessed by some object not previously known to possess them…we include this new object in the class; but it does not already belong to the class (my emphasis, VII 94–5)

12 See chapter 2, note 15.

4
The justification of deduction

1 On that definition, it becomes an inference from A to If not A then B. To prove this as a derived rule of inference requires use of the principle of reductio ad absurdum: i.e. the principle that if a premise set entails a contradiction, we may infer the negation of any one of the premises from the others. But on Mill’s view the law of contradiction, according to which every contradiction is false, is an a posteriori, inductive truth. The same point holds for the derivation of modus tollens from modus ponens: it too relies on reductio ad absurdum.

2 This is really just an extension of Mill’s points about geometrical definitions (see 5.2). The discussion in this section assumes the ‘classical pre-understanding’ of meaning: on the epistemic conception, it becomes possible to see how principles of reasoning can have a weakly a priori status even though they are not verbal. See 5.9 and 7.4.

3 Kneale and Kneale (1962:377). One can try replacing ‘things’ by ‘attributes’ and ‘coexist’ by ‘are coinstantiated’ —which is what Mill has in mind; but that makes the second axiom false, and still does not clarify the relation of the first to the syllogisms it is supposed to sustain.

The chaos is compounded in ii.ii.4, where Mill comes up with a new and single version of the supposed fundamental axiom—

whatever has any mark, has that which it is a mark of. Or, when the minor premise as well as the major is universal,… Whatever is a mark of any mark, is a mark of that which this last is a mark of. (VII 181)

4 The conclusion that it is insurmountable could be said to represent one aspect of Wittgenstein’s transition from his early to his late philosophy: from the Tractatus to the Remarks on the Foundations of Mathematics. Cf. Ramsey’s ‘Critical Notice’ of the Tractatus (Ramsey 1931:277), Michael Dummett’s The Justification of Deduction’ (in Dummett 1978) is an important discussion of the issues. (Dummett considers Mill’s position but his interpretation of Mill differs from that given here.)

5 Mill on money: Principles of Political Economy. the discussion of the Mercantile System, in ‘Preliminary Remarks’ (II 4–7), and of the ‘Purposes of a Circulating Medium’, (bk.iii, ch.vii, sec.1, III 502–3).

6 Ramsey (1978:134), ‘General Propositions and Causality’. Ramsey’s attitude in this late paper (1929) had shifted from his agreement two years earlier with the Tractatus view that general propositions should be regarded as conjunctions of their instances (‘Facts and Propositions’, p.54). In the earlier paper he explicitly remarks (p.55) that the advantage of this view is that it makes the inference from ‘For all x, fx’ to ‘fa’ tautologous. In the 1929 paper, he gives reasons for rejecting it, and goes on, ‘If then it is not a conjunction, it is not a proposition at all’, in which case, as he says, ‘the question arises in what way it can be right or wrong’.

7 Fully thought through, the analysis must lead to a dispositional account of general beliefs. Such an account was applied to beliefs in general by Bain, but not accepted by Mill—see chapter 2, note 7.

Ramsey’s thinking on generality, causation and induction breathes a thoroughly Millian air. This is less often noticed than the influence on him of Peirce and of the early Wittgenstein. He, rather than Russell, represents the classic next stage after Mill in the development of the Anglo-Saxon tradition of naturalistic empiricism. His early death cut short a rethinking of the tradition which could have become as influential as the related lines of development pioneered by the later Wittgenstein and by Quine.

8 Two senses in which deductive inference may be held to be indefeasible should be distinguished. (1) If a conclusion is a deductive consequence of certain premises, the addition of further premises cannot defeat the proof (cf. chapter 5, note 15); (2) logical principles are not revisable. A Millian empiricist cannot accept (2) in the case of those logical principles which he considers inductive truths. But he must explain in what way (1) is true.

5
Empiricism in logic and mathematics

1 Mill discusses the deductive sciences in System of Logic ii.iv-vii. The question ‘why there are deductive sciences’ is considered in ii.iv (‘Of Trains of Reasoning, and Deductive Science’).

2 The quotation is from a letter to W.G. Ward, 28 November 1859 (XV 646, no. 423). The habit of formal conservatism, and the corresponding habit of discovering ‘new’ meanings latent in old doctrines, is Coleridgean.

3 They could be taken to be regions of space, or of space-time. Treating such regions as entities is not in itself incompatible with a naturalistic philosophy—so long as they do explanatory work: that is, if there are grounds for crediting them with causal properties. (Cf. Field 1980:35–6 and note 23.)

4 Given the finitism which is implicit in Mill’s account of deductive and inductive reasoning, it might be argued that we can attach no sense to the idea of an actual object without three-dimensional extension, as distinct from the idea of a potentially infinite sequence of ever smaller extensions. But Mill does not follow up any such line of thought.

5 Mill explicitly distinguishes, in this footnote, his use of the term ‘hypothesis’ with respect to geometrical postulates from the general notion of a scientific hypothesis.

Note his claim that geometrical postulates are not ‘literally true’. That will be so only if they are taken as irreducible assertions about non-existent extensionless ‘points’ and breadthless ‘lines’. On the other hand, if such terms can be semantically analysed as limit concepts, and paraphrased out, then the postulates are literally true. There is the same alternative here as with respect to phenomenalism: do material object statements literally mean the same as statements about actual and possible sensations, or is it rather that that is all they ‘come down to’?

6 Mill gives two different axioms at VII 258: The sums of equals are equal, The differences of equals are equal’. This seems to be a slip.

7 Mill tries to establish a further point of analogy between arithmetic and geometry: that the axioms and definitions of arithmetic, and thus all its propositions, are ‘hypothetical’ in the same sense in which those of geometry are. But the comparison is strained. Mill admits that it holds only where arithmetic is applied in measuring operations, that is, in deductions concerning measured quantities: here the application of arithmetical laws requires the idealising assumption that every measured-off unit is exactly equal in magnitude. But this is a point about the application of arithmetic, not about arithmetic as such. On Mill’s own account arithmetic is not ‘hypothetical’ in the way geometry is: aggregates are not ideal-limit entities as points, lines and planes are.

8 Kessler (1980), in the course of a valuable discussion of some of Frege’s criticisms of Mill, erroneously takes this to be Mill’s actual position. Frege too (1950:23) takes Mill’s position to be ‘that the number is a property of the agglomeration of things’, though he quotes Mill as holding that the name of a number connotes that property. It should be noted that Frege had not at this stage made the distinction between sense and reference.

Remember that Mill thinks singular abstract names can be paraphrased out (2.4); so he could adopt this analysis of ‘names of number’ and simultaneously hold that all such names could be eliminated, leaving only general names—‘a three’, ‘three-membered’.

9 On this view the necessity of ‘3=2+1’ is exactly on a par with the necessity of ‘Heat is molecular motion’. The empiricist has the same problem in accounting for the apparent necessity in both cases; it is a different problem from that of accounting for the aprioricity of ‘3=2+1’.

10 ‘…it is inconceivable and impossible that, for any individuals a and b, {a, b} and {a, {a,b}} could have different spatio-temporal locations, or that they could exert different forces or undergo different changes’ (Burge 1977:103. Cf. p.114).

11 If numbers are taken to be physical properties, the question turns into one about the existence of such properties. Can a physical property exist without being instantiated? Surely not; and then the question once again turns, given that aggregates must be ‘first-order’, on the number of physical individuals.

12 Chapter vii consists of a further examination of objections and doctrines opposed to Mill’s view that ‘all deductive sciences are inductive’. It was added to the System in 1856.

13 The point of this comment was to undercut Hamilton’s Kantian claim that logic applies beyond the phenomenal to the ‘noumenal’ domain.

14 The Law of Identity’ initially appears as the principle of the self-identity of objects or ‘A thing is the same as itself’. But this principle is discarded in favour of a principle about sentences: that synonymous sentences have the same truth value— ‘Whatever is true in one form of words, is true in every other form of words which conveys the same meaning’ (IX 374).

15 An inference rule R is defeasible if there is a set of premises P which by R warrants a conclusion C, but which can be enlarged to a set P’ which does not warrant C by R. P is enlarged by adding further premises without striking any premises out. Inductive rules of reasoning are in this sense defeasible, deductive rules are not.

16 It can be found in the System, ii.v.5–6 and ii.vii.1–4; and the Examination, chapter vi.

17 It is said that Gauss attempted to test Euclidean geometry by measuring the angle-sum of the triangle formed by three distant mountain tops. The experiment might in principle have refuted Euclidean geometry, but it could not, within the limits of possible accuracy, confirm it as against Riemannian geometry.

18 Cf. Reid: ‘Experience informs us only of what has been but never of what must be’ (Woozley (ed.): 405).

19 There is a good discussion of the possibility of a ‘counter-imaginative’ arithmetic in Craig (1986: section I).

20 Spencer’s

theory of associations added little that was new to the associationism of his time; what was new was his hypothesis of the inherited effects of past associations, so that in the history of the race—and not merely in the history of each individual—more and more complex and reliable associations came to be formed. (Mandelbaum 1971: 300)

As Mandelbaum points out (p. 232) this view would not have been acceptable to earlier forms of associationism; Spencer argued for it by assuming a correlation between mental habits and physical changes in the nervous system, and arguing that habit-induced bodily changes could be inherited. The mechanism was accepted, though in a much more restricted way, by Bain (Mandelbaum 1971: 232) —a close collaborator of Mill’s, and -a leading figure in associationist psychology. Evolutionary and physiological perspectives in psychology were accepted by Mill as perfectly legitimate, but it would be fair to say that he never took their full measure in his philosophical thought.

21 Husserl (1970), chapter III, ‘Psychologism, its Arguments and its Attitude to the Usual Counter-Arguments’; chapter V, ‘Psychological Interpretations of Basic Logical Principles’.

6
Induction and inductivism

1 Ramsey (1978:100). The passage is from the 1926 article, Truth and Probability’.

2 The inference from the fact that all observed As have been Bs to the conclusion that all As are Bs is a special case of the ‘straight rule’: to infer from the fact that n% of a sample of As are Bs the conclusion that n% of all As are Bs. We shall not need to consider the more general case in what follows.

3 Cf. Putnam (1975: vol.1, The “Corroboration” of Theories’, sections 1–3). Stove (1985) is a highly unfair, but witty and penetrating, account of Popper’s philosophy and its sources.

4 The fact is well brought our in Scarre (1983).

5 The confusion is further compounded when in the 1851 edition Mill adds a qualifying footnote which is in line with his general position but completely throws into doubt his apparent acceptance of the traditional idea:

But though it is a condition of the validity of every induction that there be uniformity in the course of nature, it is not a necessary condition that the uniformity should pervade all nature. It is enough that it pervades the particular class of phenomena to which the induction relates. An induction concerning the motions of the planets, or the properties of the magnet, would not be vitiated though we were to suppose that wind and weather are the sport of chance, provided it be assumed that astronomical and magnetic phenomena are under the dominion of general laws. Otherwise the early experience of mankind would have rested on a very weak foundation; for in the infancy of science it could not be known that all phenomena are regular in their course. (VII 310)

(Cf. also the footnote added in the 1851 edition at VII 568.)

6 There are interesting points to be developed here about the way in which distinctions among spontaneous modes of reasoning, between those which are legitimate and those which are ‘superstitious’, can develop only after a certain stage in the natural history of reasoning—of the domestication of the savage mind—has been passed.

7 A single instance could never be enough for a purely enumerative induction. When he refers to a ‘complete induction’ Mill has a use of the eliminative canons in mind.

8 By a phenomenon, in this context, Mill means an observable natural process, not a subjective appearance.

9 ‘Mill’s account is a great improvement on Hume’s: he explicitly recognises a number of important complications’ (Mackie 1974:60). Mackie lucidly discusses them in his chapter 3. It is on the basis of Mill’s analysis that he develops his own notion of a cause as an ‘INUS condition’.

10 There are certain standard prima facie difficulties for this account which cannot be examined fully here. Consider the two following points.

Mackie (1974:35, 63) points out, following Anderson. (1962), the importance of the notion of a ‘causal field’. In ordinary eliminative reasoning and ascriptions of causality, the assemblages of sufficient conditions which are identified as causal antecedents are sufficient only relative to an assumed but undefined background domain of inquiry. That seems incompatible with Mill’s claim that a cause in the philosophical sense is an unconditionally sufficient cause. But the tension is only a superficial one. The regulative ideal for scientific inquiry is precisely the elimination of such relativity to an undefined background field. See also note 12.

If a cause of a phenomenon-type is an assemblage of conditions which it invariably and unconditionally follows, the proposition that causes precede their effects seems to be merely verbal. There is something wrong with this: there may be deep reasons which rule out backward causation, but its impossibility should not be an elementary verbal truth. There ought to be a way for an empiricist to avoid the conclusion that it is, but this is not the place to discuss the problem.

11 See Mill’s letter to Sir John Herschel, 1 May 1843 (XIII 583) in which he acknowledges his debt to Herschel on the matter of inductive methods. Also Mill’s comment on Bacon: by pointing out the insufficiency of enumerative induction alone, he ‘merited the title so generally awarded to him, of Founder of the Inductive Philosophy… [but the] value of his own contribution to a more philosophical theory of the subject has certainly been exaggerated…’ (VII 313).

Despite the title of iii.viii—‘Of the Four Methods of Experimental Inquiry’— Mill states five; naming them the Methods of Agreement, and of Difference, the Joint Method of Agreement and Difference, the Method of Residues and the Method of Concomitant Variations. The odd man out seems to be the Joint Method of Agreement and Difference; see the evidence given in the textual introduction to the Collected Works edition of the System (VII lxviii, note 49).

The methods have been surveyed by Mackie and extensively revised and put on a more general footing (Mackie 1974: appendix). Mackie has also discussed and neutralised a variety of objections to Mill’s conception of the eliminative methods. The present discussion follows the main points of Mackie’s account.

12 Is the fact that a certain meteor, far out in distant space, was not deflected millions of years ago in such a way as eventually to strike Kennedy’s assassin the moment before he fired, part of the cause of Kennedy’s death? Mill’s way of trying to cope with this point, i.e. by summing up the negative conditions ‘under one head, namely the absence of preventing or counteracting causes’ (VII 332), will not do, at least as an account of the ordinary notion: because we do not regard absence of counteracting causes as part of the cause. Here again the notion of a causal field is needed—the field will include some negative conditions (brake failure) but exclude others.

13 For a discussion of the issues involved, see Mackie (1974: chapter 6, ‘Functional Laws and Concomitant Variation’, and appendix, section 3, ‘Methods of Concomitant Variation’). The Method of Concomitant Variations is the basis of factor analysis.

14 Cf. Jackson (1937–8).

15 ‘Inartificial’, like ‘spontaneous’, is Mill’s word. It is obviously designed to avoid the misleading implications of ‘natural’ or ‘instinctive’: the contrast is with rule-governed, codified—domesticated.

16 Stroud (1977) suggests that Hume’s inductive scepticism can be interpreted along these lines. On this view, Hume would not be raising the straightforward question of what reason we have for accepting EI as a rule of inference. He would, rather, be accepting the rule of inference as legitimate, but pointing to a feature of it which means that we can never attain the pool of premises required to apply it. If so then his own empiricism about causation provides the basis for a reply, as suggested in the text. The reply of course still leaves open the straightforward question.

17 Mill’s views on the character of hypotheses and their use in science will be found in iii.ii (Of Inductions Improperly So-called’), iii.xi (‘Of the Deductive Method’) and iii.xiv (^‘Of the Limits to the Explanation of Laws of Nature; and of Hypotheses’).

18 My object was to analyse, as far as I could, the method by which scientific discoverics have really been made; and I called this method Induction…. That is not exactly the Induction of Aristotle, I know, nor is it that described by Bacon…. I am disposed to call it Discoverer’s Induction; but I dare not venture on such a novelty, except in the indirect way in which I have done. (Whewell to de Morgan, 18 January 1859. In Todhunter 1876: II, 416. The passage is quoted in Tewari 1980) Notice that in this disagreement about the status of the Hypothetical

Method, Popper in fact stands on the side of Mill, not Whewell. Like Mill he thinks that the Hypothetical Method cannot be regarded as an independent way of justifying belief. Unlike Mill, of course, he thinks that is true of any inductive reasoning.

Peirce, in contrast, stands on the side of Whewell. The underlying issue between Mill and Popper and Whewell and Peirce is that of metaphysical realism. We take it up in 7.4.

19 The impossibility of arriving at the right description by direct observation is, Mill thinks, a contingent limitation: if the planets left visible tracks, an observer suitably placed in space could directly perceive them to be elliptical. There are important weaknesses in Mill’s discussion of this example (iii.ii.3–5), the main one being the false contrast he draws between ‘description’ and ‘induction’. Kepler’s first law may in an intelligible sense be termed descriptive rather than explanatory; but that does not mean that it was not arrived at by inductive inference, nor does Mill really deny that it was.

The point Mill wants to resist is Wheweil’s idealism-tending claim that hypotheses introduce ‘a conception of the mind, which did not exist in the facts themselves’ (quoted by Mill, VII 294). Whewell ‘expresses himself as if Kepler had put something into the facts by his mode of conceiving them. But Kepler did no such thing… Kepler did not put what he had conceived into the facts, but saw it in them’ (VII 295). This cries out for restatement. Obviously Kepler neither put the planetary ellipse in the heavens nor saw it there. But Mill’s polemic always deteriorates when he feels under pressure, and he seems to have felt that way whenever he argued with Whewell.

7
Induction, perception and consciousness

1 The terms ‘manifest image’ and ‘scientific image’ are taken from Wilfrid Sellars (1963).

2 Mill discusses the term ‘positivism’ in Auguste Comte and Positivism (e.g. X 263–7). For the general nineteenth-century agreement on the phenomenal relativity of knowledge see Mandelbaum (1971). The decline of the ‘method of hypothesis’ in the eighteenth century, and its resurgence in the nineteenth, is discussed in Laudan (1981).

3 ‘Whewell’s critics often pointed out his failure to perceive that something like Kant’s “transcendental deduction” was required to justify the “Fundamental Ideas” as the conditions of all necessary truths’ (Tewari 1980: 114).

The quotations from Whewell are from his article ‘On the Fundamental Antithesis of Philosophy’, printed as Appendix E in Whewell (1860). The aphorism in full is ‘Man’s intellectual progress consists in the Idealization of Facts, and his Moral Progress consists in the realization of Ideas’.

4 If Whewell

reduces the undulations to a figure of speech, and the undulatory theory to the proposition which all must admit, that the transmission of light takes place according to laws which present a very striking and remarkable agreement with those of undulations… Ive no difference with him on the subject. (VII 504, footnote)

5 Michael Dummett has made a penetrating exploration of the contrast between the classical pre-understanding of meaning and the epistemic conception of it; or, to use his terms, between ‘realism’ and ‘anti-realism’. See especially Dummett (1978).

6 The positions of the natural realist and the scientific realist—who both in their own way think that something can be known about things as they really are—seem to be ignored in this dichotomy. But Mill is accurately reflecting the nineteenth-century consensus, in which pre-Kantian versions of these forms of realism were no longer possible. Hamilton is a good example of this: he does not simply endorse Reid’s natural realism in a pre-Kantian way. The alternative to phenomenalism was an empirical realism (natural or scientific) which stepped the unknowability of things-in-themselves up to a transcendental level. Thus Mill places Kant in the second school. A Kantian can, he recognises, accept the scientific realist’s distinction between primary and secondary qualities, but only in a way which respects the doctrine of relativity—the unknowability of things as they really are:

Such properties as the objects can be conceived divested of, such as sweetness or sourness, hardness or softness, hotness or coldness, whiteness, redness or blackness—these it is sometimes admitted, exist in our sensations only. But the attributes of filling space, and occupying a portion of time, are not properties of our sensations in their crude state, neither, again, are they properties of the objects, nor is there in the objects any prototype of them. They result from the nature and structure of the Mind itself: which is so constituted that it cannot take any impressions from objects except in those particular modes. We see a thing in a place, not because the Noumenon, the Thing in itself, is in any place, but because it is the law of our perceptive faculty that we must see as in some place, whatever we see at all…. Time and Space are only modes of our perception, not modes of existence, and higher Intelligences are possibly not bound by them…. (IX 9)

We return to the relationship between Mill’s phenomenalism and transcendental idealism in 7.9.

7 In fact Reid does not always use ‘perceive’ to signify a relation between subject and external objects. He also uses the term in such a way that the mere occurrence of a sensation which gives rise to appropriate beliefs counts as an ‘act of perception’. Perceptions can therefore be erroneous:

Nature has connected our perception of external objects with certain sensations. If the sensation is produced, the corresponding perception follows, even where there is no object, and in that case is apt to deceive us. (Reid 1846:315; Mill, IX 173)

Compare Hamilton on Reid:

It is palpably impossible that we can be conscious of an act [of perception] without being conscious of the object to which that act is relative. (Quoted by Mill, IX 111)

An act of knowledge existing and being what it is only by relation to its object, it is manifest that the act can be known only through the object to which it is correlative; and Reid’s supposition that an operation can be known in consciousness to the exclusion of its object, is impossible. For example, I see the inkstand…. Annihilate the inkstand, you annihilate the perception; annihilate the consciousness of the object you annihilate the consciousness of the operation. (IX 111)

8 There is a remark which may seem incompatible with this in one of Mill’s letters to W.G. Ward (No.423, 28 November 1859, XV 648):

Memory I take to be the present consciousness of a past sensation. It is strange that such consciousness can exist; but the facts denoted by was, is, & is to come, are perhaps the most mysterious part of our mysterious existence, as is strikingly expressed in the well known saying of St Augustine.

However it is not clear whether the phrase ‘present consciousness of a past sensation’ is meant to imply that there is no present ‘memory-datum’ which I am aware of in having that consciousness.

9 The relevant sections of the Examination are chapters xi-xiii; including the important appendix to chapters xi and xii.

The term ‘phenomenalism’ was not used by Mill, though W.G. Ward described Mill’s position as ‘phenomenism’. It could appropriately be used, as it is by Maurice Mandelbaum, to refer to the doctrine that all knowledge is phenomenal, in the nineteenth-century sense; but here I am using it in its usual current sense.

10 No. 863, 12 August 1865, Mill is referring to Spencer (1865)—and the argument he is addresssing is an old and influential one. (It is approved by Mandelbaum: see Mandelbaum 1971:496, note 43). Spencer points out that arguments for subjective phenomenal relativity rest on ‘physicalistic’ or ‘objectivist’ premises. But this does not show, contra Spencer and Mandelbaum, that phenomenalism is untenable. On the contrary, if these arguments are valid, then they constitute a reductio of the objectivist position, or at least its knowability.

11 Bradley (1927:39–40, footnote) sums it up with a characteristically grating sneer:

with the…fact before him, which gave the lie to his whole psychological theory, he could not ignore it, he could not recognise it, he would not call it a fiction; so he put it aside as a ‘final inexplicability’, and thought, I suppose, that by covering it with a phrase he got rid of its existence.

12 There are of course other complex issues in the offing. Reflection on our criteria of continuing personal identity across time may force us to recognise a fictional element in our notion of persons. An eliminative materialist view of mental states would force on us the conclusion that there is no subjective bearer of these non-existent states. But these lines of thought should be distinguished from the matter at issue in Mill’s discussion—the relativity of our knowledge of the self.

13 A further question is whether the ‘series’ view of the self involves denial of active intelligence. This is Alan Ryan’s view:

Mill’s philosophy required an active mind which would construct an external world out of sensations, and order it according to rationally organized theories; and yet he had no way of accounting for the existence of such an active intelligence. If the external world was to be constructed out of experience by a self which tried out inductive hypotheses about the course of its experience, then this presupposed a unitary self to do the experiencing, and to make the inferences. Yet the atomistic theory to which Mill was attached seemed to rule out any such self. This means that the metaphysics to which Mill was committed had a contradiction at its heart. (Ryan 1974:226)

But the contradiction is at least not obvious. It is true that Mill’s psychology can make no sense of the idea of an irreducible ‘agent causality’. (Mill on agent causality: IX 441, footnote.) But who can? If recognising agent causality is a precondition of making intelligible the notion of a rational knower and agent, then that concept itself is in bad shape. The problems here are posed by naturalism as such, not by this or that version of it. See 8.2 and 8.10.

14 I take the term ‘plain’ from Clarke (1972).

15 For the reason given by Descartes in Meditation VI and restated forcefully and illuminatingly by Kripke (1971; 1980). See also Nagel (1979).

The central issue is not what the essential properties of the ‘material’ or ‘physical’ are. The point is that pure experience cannot be identified with anything that can be characterised objectively. This gives us a fresh description of naturalism: it is the view that human beings can be fully characterised objectively.

8
The logic of the moral sciences

1 From the epigraph of Book vi of the System of Logic, taken from Condorcet’s Esquisse d’un tableau historique des progrès de I’esprit humain.

2 In the Examination, chapter xxvi, ‘On the Freedom of the Will’, Mill rightly emphasises that we could not have a direct self-knowledge of fireedom: ‘Consciousness tells me what I do or feel. But what I am able to do, is not a subject of consciousness’ (IX 449). He notes the ‘pretended inconsistency’ alleged by a commentator as between this statement and the references in the System to a ‘practical feeling of Free Will’ and ‘a feeling of Moral Freedom which we are conscious of’, and comments:

When I applied the words feeling and consciousness to this acquired knowledge, I did not use those terms in their strict psychological meaning, there being no necessity for doing so in that place; but, agreeably to popular usage, extended them to (what there is no appropriate scientific name for) the whole of our familiar and intimate knowledge concerning ourselves. (IX 450, note)

3 Relevant chapters in the System: iii, vi: ‘Of the Composition of Causes’; xii-xiii: ‘Of the Explanation of Laws of Nature’, ‘Miscellaneous Examples of the Explanation of Laws of Nature’; xvi: ‘Of Empirical Laws’; xxiii: Of Approximate Generalizations, and Probable Evidence’.

4 Mill’s several references to Dalton’s theory (VII 221, 375, 473) show a marked reluctance to accept it as evidencing an underlying atomic reality. He refers to ‘the principle of Dalton, called the atomic theory, or the doctrine of chemical equivalents’ (VII 221) and again to ‘Dalton’s generalization, commonly known as the atomic theory’ which established a ‘table of the equivalent numbers, or, as they are called, atomic weights, of all the elementary substances’ (VII 473–4).

5 Mill takes it for granted that if the phenomena of mind can be shown to arise from physical laws, it will be by chemical combination not mechanical composition, and that is because he does not consider the possibility that they may be discovered to be physical phenomena:

To whatever degree we might imagine our knowledge of the properties of the several ingredients of a living body to be extended and perfected, it is certain that no mere summing up of the separate actions of those elements will ever amount to the action of the living body itself. The tongue, for instance, is, like all other parts of the animal frame, composed of gelatine, fibrin, and other products of the chemistry of digestion, but from no knowledge of the properties of those substances could we ever predict that it could taste, unless gelatine or fibrin could themselves taste; for no elementary fact can be in the conclusion, which was not in the premises. (VII 371–2)

The last sentence shows up clearly the lacuna in Mill’s account of scientific reduction. It is true that a sentence containing the term ‘taste’ cannot be deduced from sentences not containing it (or a synonym for it). But the moral has to be drawn with care. What if tasting something sour, say, can be identified with a particular physical process in the central nervous system? The premises from which the sentence ‘It tastes sour’ is deduced will then still contain the phrase ‘rastes sour’, because they will contain a theoretical identity which identifies ‘tastes sour’ with ‘stimulates activity in such and such neural fibres’. But that will not show that there are irreducible laws of taste in the sense envisaged by Mill.

On the other hand, if states of consciousness cannot be strictly identified with physical states then the mental will indeed be emergent, and its laws, if any, heteropathic.

The term ‘materialism’ was used by Mill and more generally in the nineteenth century to refer to epiphenomenalism. See Mandelbaum (1971: 21).

6 He never links the issue with his phenomenalism. No doubt he would insist that psychophysical laws must be further reducible to causal laws holding between sensations and permanent possibilities of sensation.

7 This not a complete list of associationist principles. Associationism is essentially a theory of concept-formation. There is a further question about how these concepts become contents of intentional attitudes. One aspect of this is the problem of how associationism generates belief (see chapter 2, note 7). Another is how it generates desires, emotions or volitions. There have to be laws postulating natural or innate feelings, and then explaining how ‘affect’ is transferred from the conceptual content of these to new conceptual contents.

8 There is a reply to Comte’s criticism of the method of introspection in Auguste Comte and Positivism.

The importance of associationism for Mill’s belief in the malleability and progressiveness of man has been stressed by Maurice Mandelbaum. (In itself, of course, associationism need not entail a doctrine of inherited cognitive and emotional equality among human beings. Bain thought Mill’s desire to believe such a doctrine his greatest error as a ‘scientific thinker’: Bain (1882:146).)

9 The main texts in this controversy are collected in Lively and Rees (1978), which also contains a useful introduction.

10 Weber’s example comes from Weber (1975:171). I have altered the wording. The view of interpretation as phronesis is interestingly developed in Gadamer (1979). Gadamer also quotes a marginal comment from Dilthey’s copy of Mill’s System: ‘Only in Germany could the practice of an authentic experience be substituted for an empiricism which was dogmatic and burgeoning with prejudices; Mill is dogmatic for lack of historical erudition’ (Gadamer 1979:118). It would be a lengthy but worthwhile business to disentangle what is right and what is wrong in that.

11 Donald Davidson has brought out clearly the relevance of such examples. See, for example, Davidson (1978:153ff-).

12 There is a discussion of the idea of ‘tracking bestness’ and its relation to the issue of free will and determinism in Nozick (1981: chapter 4, section II).

13 Dennett (1973). See also the whole of Dennett (1977). Much fascinating work is currently being done on these matters in the frontier area between psychology and philosophy.

9
Utilitarianism

1 ‘Mill has made as naive and artless a use of the naturalistic fallacy as anybody could desire’, etc. (Moore 1948:66).

On 18 March 1868, Mill’s German friend and translator wrote to him about the translation of Utilitarianism he was preparing:

Let me conclude by expressing my regret that you did not in the later editions of the Utilitarianism remove the stumbling block (to any reader and more especially to the translator)… (audible, visible-desirable) which when pointed out to you by me, you said you would remove. Your argument looks like a verbal quibble, far as it is from being one and has besides to me the serious disadvantage of being utterly untranslatable. (Quoted at X cxxvi)

Mill replied on 23 April 1868:

With regard to the passage you mention in the Utilitarianism I have not had time regularly to rewrite the book & it had escaped my memory that you thought that argument apparently though not really fallacious which proves to me the necessity of, at least, further explanation & development. I beg that in the translation you will kindly reserve the passage to yourself, & please remove the stumbling block, by expressing the real argument in such terms as you think will express it best. (XVI 1391)

It is curious to compare the small importance Mill attached to his formulation with the mountain of literature it has produced.

2 The example of the loving father comes from Peter Winch’s ‘Moral Integrity’, in Winch (1972). Kantian obsession with conscientiousness arises from the idea that what is done from inclination must be done for one’s own sake, and hence cannot have moral worth. Neither can an act done from inclination be authentically autonomous. So for the Kantian the categories: acting from inclination, acting selfishly and acting heteronomously, collapse into each other. It is an important feature of the humanistic tradition of ethical thought to which Mill belongs that it does not collapse them.

The similarity between Butler’s discussion of self-interest, and Mill’s, is pointed out by Maurice Mandelbaum (1968a: 39). Cf. also Mill’s remarks on the hedonistic fallacy, cited in chapter 1, note 5.

An interpretation of Mill’s ‘proof’ which attempts to read it in accordance with the views on pleasure and desire Mill expressed in his early article on Bentham is given by Berger in his excellent study of Mill’s moral and political philosophy. I have criticised this approach in my review, to which Berger has replied (Berger 1985).

3 For this reason among others—i.e. the possible gap between what I am justified in holding to be true, on any feasible improvement of my information, and what is true—the conception of well-being proposed by James Griffin, in which well-being is conceived as ‘the fulfillment of informed desire’ (Griffin 1986:75), cannot be right.

It will not help to define informed desire as what I would desire if I believed all relevant truths and formed rational preferences in response. In the first place the concept of informed desire is then doing no real work. Second, to get the right results about what my actual utility schedule is, such an account might well have to equip me with information about what my actual beliefs are together with information showing these beliefs to be false—but that could well cause psychological changes producing a changed utility schedule. See also Hurka (1988: part I).

But though the conception of well-being developed here is not an ‘informed-desire’ account, it does not blankly lay down an ‘objective list’ of ends either; categorial ends are so because desired in theory and in practice, that being established by the method of reflection and self-examination Mill uses. Of course the relative importance of categorial ends for a particular person is a matter of individual character, discoverable only by experience— on this as on many other points (e.g. about incommensurability, and about the foundations of liberalism) I am in agreement with Griffin.

4 In chapter XIX of the Analysis of the Phenomena of the Human Mind, James

Mill says:

The term ‘Idea of a pleasure,’ expresses precisely the same thing as the term, Desire. It does so by the very import of the words. The idea of a pleasure, is the idea of something as good to have. But what is a desire, other than the idea of something good to have? (Mill 1869:191–2)

A little later he qualifies this.

The idea of a pleasurable sensation with the association of the Past, is never called Desire. The word Desire, is commonly used to mark the idea of a pleasurable sensation, when the future is associated with it. The idea of a pleasurable sensation, to come, is what is commonly meant by Desire. (193)

J.S.Mill added an editorial comment to the chapter in which he comments that ‘Desire is not Expectation, but is more than the idea of the pleasure desired, being, in truth, the initiatory stage of Will’ (104). In other words, Mill is prepared to grant that one might think of something as pleasant without having the disposition to act to get it. But he does not disagree with the points—vital as far as his ‘proof’ of hedonism is concerned—that (1) the idea of pleasure is identical with the idea of ‘something as good to have’, and (2) that desiring something at least necessarily involves thinking of it as ‘good to have’. The second point is true, but the first is false.

On Mill’s use of the word ‘metaphysical’, as in the passage in the text, to mean ‘psychological’, see Mandelbaum (1968a: 39).

5 Early criticism of this type is documented in Schneewind (1977: chapter 5, sec.v, ‘Mill’s Utilitarianism and its Reception’, esp. 185–6).

6 Mackie (1976b). For more on Sidgwick’s argument see also Schneewind (1977) and Skorupski (1979). I take the distinction between agent-neutral and agent-relative reasons, but not the terms, from Nagel (1970). (He calls them ‘objective’ and ‘subjective’ reasons; chapter X.)

7 In particular, the ‘lexical’ version of Rawls’ difference principle is compatible with it:

first maximise the welfare of the worst-off representative man; second, for equal welfare of the worst-off representative, maximise the welfare of the second worst-off representative man, and so on until the last case which is, for equal welfare of all the preceding n-1 representatives, maximise the welfare of the best-off representative man. (Rawls 1972:83) Cf. Gauthier (1982).

8 The interconnections between prudence, utilitarianism and a ‘reductionist’ view of persons are very fully explored in Parfit (1984: chapters 14 and 15).

The unity and distinctness of an individual life are obviously both central to the liberal vision, via notions involved in auconomy such as a plan of life and a private domain. (See 9–4, 10.5, 10.8.) But it is far from clear that they are undermined by a rejection of metaphysical notions of self-identity. For a liberalism based on utility, they would be undermined only if such rejection in practice caused a reduction in the desire for autonomy; a reduction in its importance in relation to such categorial ends as happiness or knowledge. It is not obvious that it would or should.

9 Mill’s indirect utilitarianism is discussed in Mandelbaum (1968a). John Gray summarises the reasons why Mill should not be classed as either an ‘act’ or a ‘rule’ utilitarian, and defends Mill’s indirect utilitarianism against objections (chapter II, section 2, ‘Acts, Rules and the Art of Life’). The clearest recent discussion is Berger (1984: chapter 3, The Greatest Happiness Principle and Moral Rules’).

10 Other reasons: mistaken do-gooders, problems of co-ordination etc. There is also a problem about how to state the act-utilitarian position without lapsing into incoherence. Acquiring information and making calculations are acts, as, strictly, is taking one’s beliefs for granted. The cost-effectiveness of these acts themselves could therefore be calculated. But that again is an act, whose cost-effectiveness could be calculated…. This is the ‘paradox of information’. Not that I mean to suggest that act-utilitarianism can be refuted by being shown to be incoherent. We know what is meant by it.

11 The section of Grote’s letter to which the quoted passages are a reply unfortunately seems not to have been preserved. Relevant comments are found throughout Mill’s letters (e.g. XVI 1254, 1327). The following passages from ‘Taylor’s Statesman’, which Mill co-authored with George Grote, are noteworthy:

To admit the balance of consequences as a test of right and wrong, necessarily implies the possibility of exceptions to any derivative rule of morality which may be deduced from that test…. The evil of departing from a well-known and salutary rule is indeed one momentous item on that side of the account; but to treat it as equal to infinity, and as necessarily superseding the measurement of any finite quantities of evil on the opposite side, appears to us to be the most fatal of mistakes in ethical theory….

moral rules are perpetually liable to clash with one another, and actually do so clash in all those exceptional cases now under consideration, so as to leave us no resource except in a direct appeal to the supreme authority from whence all moral rules are derived…. (XIX 638–40)

12 The balance is nicely struck in the following passage from Utilitarianism;

There is no difficulty in proving any ethical standard whatever to work ill, if we suppose universal idiocy to be conjoined with it; but on any hypothesis short of that, mankind must by this time have acquired positive beliefs as to the effects of some actions on their happiness; and the beliefs which have thus come down are the rules of morality for the multitude, and for the philosopher until he has succeeded in finding better. That philosophers might easily do this, even now, on many subjects; that the received code of ethics is by no means of divine right; and that mankind have still much to learn as to the effects of actions on the general happiness I admit, or rather, earnestly maintain. (UII24, X 224)

13 I say ‘rational kernel’, because the unreconstructed concept of a right presents it as something which a person possesses and in virtue of which an obligation exists—which is not reducible to the obligation but on the contrary explains its existence. This dimension is lost in Mill’s analysis. Nor is it possible to retrieve it. The ‘explanatory’ dimension in concept of a moral right really does dissolve when pressed, in the way in which Mill (wrongly, in this latter case) thinks that the explanatory dimension in the concept of a physical object dissolves when pressed (see 10.3).

Berger (1984) examines Mill’s application of the notion of justice and rights across a wide span of substantive political issues. (He comments that he knows ‘of no place outside of [Utilitarianism] where Mill explicitly invoked his theory of rights’ (Berger 1984:191). There is an interesting, though brief, discussion of the concept of a moral right in ‘Use and Abuse of Political Terms’ (1832), XVIII 3–13; Mill comments, ‘Right is the correlative of duty, or obligation; and (with some limitations) is co-extensive with those terms. Whatever any man is under an obligation to give you, or do for you, to that you have a right’ (XVIII 8). This is cruder than Utilitarianism, where the analysis is considered enough to be called a theory, but on the same basic lines.)

14 He considers such a case in a letter to William Thomas Thornton of 17 April 1863 (XV 853–4). Mill first refers to the example Thornton had given, of Agamemnon and Iphigenia. He then introduces an example of his own: should the Carthaginians have surrendered Hannibal to the Romans in order to safeguard the security of the city? He thinks ‘there can be no doubt that the morality of utility requires that the people should fight to the last rather than comply with the demand’ —because of the indirect utility of resisting tyrannous aggression.

15 In reality things are more complicated of course (they always are)— unanimity is the exceptional case. Typically some will want to join a new practice and others will not. If you, coolly rational, want to join the survival club, but don’t have enough fellow-members, do I have some obligation to you to rethink my primitive aversions and consider joining it too?

16 Mill’s views on punishment and desert are considered by Berger (1984: 134–46). Interesting comments on punishment are to be found in an 1834 review by Mill, ‘On Punishment’ (XXI 73–9):

You do not punish one person in order that another may be deterred. The other is deterred, not by the punishment of the first, but by the expectation of being punished himself: and as the punishment you threaten him with, would have no effect on his conduct, unless he believed that it would really be inflicted, you are obliged to prove the reality of your intention, by keeping your word whenever either he, or any other person, disregards your prohibition….

The only right by which society is warranted in inflicting any pain upon any human creature, is the right of self-defence…our right to punish, is a branch of the universal right of self-defence; and it is a mere subtlety to set up any distinction between them…. (78)

10
Liberty

1 Morley: quoted in XVIII lxix, from Recollections, I, 61. Fox: Caroline Fox, Memories of Old Friends, p. 347, quoted in Alexander (1965:129). Carlyle: New Letters of Thomas Carlyle, II, 196, quoted in Packe (1954:405).

2 There is an interesting difference of tone between this retrospective statement in the Autobiography and the note of urgent relevance to the times which is characteristic of the essay itself. Mill may have been responding to his early critics who almost universally felt that he had greatly exaggerated the degree to which unconventional opinions were under pressure. See Rees (1985: chapter 3, ‘On Liberty and its Early Critics’).

3 Madame de Staël contrasted ‘civil’ and ‘political’ liberty:

Political liberty is to civil liberty as the guarantee is to the end for which it stands, security; it is the means and not the end and what contributed especially to make the French Revolution so disorderly was the displacement of ideas which took place in this respect. They wanted political liberty at the expense of civil liberty…. Political liberty is of consequence to ambitious men who desire power. Civil liberty interests peaceful men who only do not want to be dominated. (From ‘Reflexions sur la paix intérieure’, quoted in Dodge 1980:40)

Her lover Benjamin Constant expressed the same distinction in his contrast between the ‘liberty of the ancients’ and the ‘liberty of the moderns’ (from ‘De la liberté des anciens comparée a celle des modernes’ quoted in Dodge 1980:38–9). Liberty, to moderns,

means for every one to be under the dominion of nothing but the law, not to be arrested, detained, or put to death, nor maltreated in any way as a consequence of the arbitrary will of one or more individuals. It is for every one to have the right to express his opinion, to choose and exercise his occupation, to dispose of his property and even to abuse it, to go and come without having to obtain permission, and without having to give an accounting of his motives or actions. It is the right of each person to associate with other individuals, either to discuss their interests, or to practice the form of worship they prefer, or simply to fiil the days and hours in a way which best suits their inclinations and fancies.

4 Mill was always conscious of the potential harms of free discussion: compare the passages from the 1842 review, ‘Bailey on Berkeley’, quoted in 1.6, p. 37. It is, I think, true that he placed greatest weight on these dangers in the period when he was most influenced by Coleridge and Comte; on the other hand, there was no time when he did not come down on the side of free expression, and the dangers are still recognised and balanced in the essay On Liberty. The ‘two Mills’ thesis is put forward by Gertrude Himmelfarb (1974); see Ten (1980:151–66) for a rejoinder.

5 The phrase is de Tocqueville’s. Not that Mill was opposing democracy: as he remarks in ‘De Tocqueville on Democracy in America’ (1840) (XVIII 156),

This phrase was forthwith adopted into the Conservative dialect, and trumpeted by Sir Robert Peel in his Tamworth oration…. And we believe it has since been the opinion of country gentlemen that M. de Tocqueville is one of the pillars of Conservatism, and his book a definitive demolition of America and of Democracy. The error has done more good than the truth would perhaps have done: since the result is, that the English public now know and read the first philosophical book ever written on Democracy, as it manifests itself in modern society.

6 Ten (1980: chapter 4, section II, ‘Mill’s Concept of Harm’, p. 55). I should add that Ten generally sets an admirable standard of interpretative common sense.

7 I follow Ten (1980: chapter 2) in taking it that the Liberty Principle functions to test the eligibility of reasons for interfering with an act, so that the question becomes one of finding criteria which qualify some reasons for interference as eligible for consideration, and disqualify others. That elegantly sidesteps well-worn, and often factitious, controversies (see Rees 1960; 1985) about the possibility of distinguishing ‘self’ and ‘other-regarding’ acts.

This is not to deny however that the latter distinction is present in Mill’s thinking. Nor is it to deny its relevance, or that of cognate intuitive notions such as the notion of an individual’s private domain, or of what is his business and no one else’s. We shall appeal to these in discussing autonomy, but that is consistent with holding that the formulation in terms of reasons is basic to the interpretation of a principle for political and social liberty.

8 For Mill’s definition of ‘philosophical radical’ see ‘Fonblanque’s England under Seven Administrations’ (1837) (VI 353), where he compares ‘Philosophic’, ‘Metaphysical’ and ‘Historical’ Radicalism.

9 Arnold 1965: vol. 5 (Culture and Anarchy}, 176–7. To a considerable extent Culture and Anarchy is a response to On Liberty; Arnold initially regardcd ‘Millism’ as a debased form of Hellenism, though he later came to recognise a greater affinity between his own position and Mill’s than he had at first perceived. There is a perceptive comparison of Arnold’s and Mill’s ideals of self-development in Mandelbaum (1971: chapter 11, section 1).

10 This was a common theme among Mill’s earliest, as well as among his later idealist, critics. See Rees (1985:86ff.).

11 The contrast in this respect between Arnold and Mill is brought out in detail by Alexander (1965), especially in his last two chapters, The Best that is Known and Thought in the World’ and ‘Culture and Liberty’.

12 Two excellent guides for pursuing these fundamental questions in Mill’s thought further are Duncan (1973) and Ryan (1984).

13 The love of liberty, in the only proper sense of that word, is unselfish; it places no one in a position of hostility to the good of his fellow-creatures; all alike may be free, and the freedom has no solid security but in the equal freedom of the rest. The appetite for power is, on the contrary, essentially selfish; for all cannot have power; the power of one is power over others, who not only do not share in his elevation, but whose depression is the foundation on which it is raised. (‘Centralisation’, IX 610)

Mill’s view of the relative importance of this human want is interestingly set out in the Principles of Political Economy (11:208), also Winch (360); in a discussion of the prospects of communism:

After the means of subsistence are assured, the next in strength of the personal wants of human beings is liberty; and (unlike the physical wants, which as civilisation advances become more moderate and more amenable to control) it increases instead of diminishing in intensity, as the intelligence and the moral faculties are more developed. The perfection both of social arrangements and of practical morality would be, to secure to all persons complete independence and freedom of action, subject to no restriction but that of not doing injury to others: and the education which taught or the social institutions which required them to exchange the control of their own actions for any amount of comfort or affluence, or to renounce liberty for the sake of equality, would deprive them of one of the most elevated characteristics of human nature.

If the desire for liberty increases with affluence and moral education, there is utilitarian ground for ‘priority of liberty’ in advanced societies. (Note that Mill seems to envisage two effects: a utility trade-off between material comfort and liberty, and also a shift in the indifference curve in favour of liberty, resulting from education. Note also the perfectionist appeal in the last sentence.)

14 So a paternalist restriction on the validity of certain self-injurious contracts, as in Mill’s extreme example of a person who freely sells himself into slavery (LV11, XVIII 299), would be justified: liberty is not an unqualified side-constraint.

15 Cf. Mill LIV4–6, on the obligation human beings owe to each other ‘to distinguish the better from the worse’. The ‘self-regarding virtues’ should be encouraged by conviction and persuasion, not compulsion. ‘I do not mean’, Mill goes on,

that the feelings with which a person is regarded by others, ought not to be in any way affected by his self-regarding qualities or deficiencies. If he is eminent in any of the qualities which conduce to his own good, he is, so far, a proper object of admiration. He is so much the nearer to the ideal perfection of human nature. If he is grossly deficient in those qualities, a sentiment the opposite of admiration will follow. (XVIII 278)

The ‘self-regarding virtues’ should not be assumed co-extensive with general ideals of character, but the same points apply.

Incidentally it might be thought that ‘self-regarding’ virtues cannot be strictly moral virtues, given Mill’s analysis of strictly moral obligation (9–11), and his insistence that the Liberty Principle applies to social as well as legal compulsion. I believe this to be broadly right (waiving further matters about broader and narrower meanings of ‘moral’) —but the issue cannot be followed through here. See, for example, Berger (1984), Brown (1972), Copp (1979).

Ronald Dworkin provides an extreme example of the identification of liberalism with neutrality between moral positions when he suggests that liberal political theory

supposes that political decisions must be so far as possible, independent of any particular conception of the good life, of what gives value to life. Since the citizens of a society differ in their conceptions the government does not treat them as equals if it prefers one conception to another. (Dworkin 1978:127)

In so far as the sphere of morality is distinguished from the sphere of ideals of character there is an ingredient of truth in this (Cf. Strawson 1961). But overall it seems to me a very misleading description, at least of classical liberalism as represented by Mill. Liberalism of that kind is, on the contrary, founded on a very clear conception of what gives value to life, and would have no truck with the bizarre (or, alternatively, truistic) idea that if we prefer one man’s conception to another’s we are not treating them as ‘equals’.

16 A community as well as an individual can be said to have a private domain. I am discussing the relations between a community and one of its members. Different principles govern the relations between a community and a visitor to it. The visitor may have an obligation to respect its customs which a dissenting member may not have.

17 In Kant’s words:

The experiment…whereby we test upon the understanding of others whether those grounds of the judgement which are valid for us have the same effect on the reason of others as on our own, is a means, although only a subjective means, not indeed of producing conviction, but of detecting any merely private validity in the judgement…. (Kant 1929: A821, B849)

I consider the underlying relationships between convergence and correspondence further in Skorupski (1986b).

18 I heard a version of this argument discussed in a talk at Cambridge by Saul Kripke. Kripke presented it in terms of knowledge, not justified belief. In that version it is much harder to deal with; bringing out real tensions between the concept of knowledge and fallibilism. The application to Mill is of course my own.

19 Transcendental’: there are evident similarities between the position which emerges when the connection between fallibilism and liberty of expression is pressed, and Jürgen Habermas’s notion of the ‘ideal speech situation’. (My use of the word ‘dialogue’ is also intended to bring some of these similarities to mind.) See also 1.6 and 1.7.

20 Mill sees the tension between this dialectical doctrine of understanding and the growth of knowledge:

Is the absence of unanimity an indispensable condition of true knowledge? Is it necessary that some part of mankind should persist in error, to enable any to realize the truth? Does a belief cease to be real and vital as soon as it is generally received—and is a proposition never thoroughly understood and felt unless some doubt of it remains?… The cessation, on one question after another, of serious controversy, is one of the necessary incidents of the consolidation of opinion; a consolidation as salutary in the case of true opinions, as it is dangerous and noxious when the opinions are erroneous. But though this gradual narrowing of the bounds of diversity of opinion is necessary…we are not therefore obliged to conclude that all its consequences must be beneficial. (LII31, 32; XVIII 250–1)

There is an absence of positivist utopianism in Mill’s thought, a feeling for the disadvantages of consensus and uniformity. But does requiring ‘Socratic dialectics’ cater for the point Mill himself makes, that criticism must come from those who genuinely disagree?

21 But neither truly universal one-person one-vote suffrage, nor individual property rights, are among these ground-rules. Mill’s complex and changing views on democracy and on property and socialism are unfortunately beyond the scope of this book; they can be found mainly in Considerations on Representative Government, Principles of Political Economy and Chapters on Socialism. See also Ryan (1974; 1984) and Gray (1979).

22 David Marquand describes two conceptions of politics and social change, the ‘communal’ and the ‘exchange’ mode: ‘Society is either a kind of hierarchy, held together because those at the bottom obey those at the top, or it is a kind of market, held together by the calculating self-interest of its members’ (Marquand 1987:251). He sets against them a third mode, the ‘preceptoral’, ‘persuasive’, ‘educational’ or ‘moral’, quoting Mill—‘We do not learn to read or write, to ride or swim, by merely being told how to do it, but by doing it’ (XVIII 63, and Himmelfarb (ed.) 1963:186).

Certainly the third mode captures Mill’s view of the relationship between democracy and intellectual and moral authority: moral authority has a preceptoral role, but it can play it by persuasion—so long as background education, free institutions and access to political responsibility produce citizens who can make rational decisions because they have been exercised in making rational decisions. So this mode assumes, as Mill does, that cohesion can be maintained without appeal to non-rational bases of authority and allegiance such as faith, prejudice or deference to descent.

23 He was willing to load the dice in favour of it in practice, by means of plural voting for the educated—but never by allowing it to control the flow of expression, information and debate.