DRAFT as at 15 Nov 99

please do not quote without consulting comments appreciated

Democratic Deliberation Within

ROBERT E. GOODIN
Philosophy Program Research School of Social Sciences Australian National University Canberra ACT 0200 Australia <goodinb@coombs.anu.edu.au>

Paper presented to Conference on "Deliberating about Deliberative Democracy" School of Law, University of Texas, Austin 4-6 February 2000

1999 Robert E. Goodin

2 Democratic Deliberation Within

I. A Conceptual Prelude

The phrase 'deliberative democracy' rotates revealingly on itsaxis. Expressible alternatively as 'deliberative democracy' or as'democratic deliberation', each formulation reveals importantly differentfeatures of that ideal. Consider 'deliberative democracy'. There, 'deliberative' modifies what sort of democracyit is we want. To say that democratic decision-making ought be'deliberative' implies, first of all, that it be 'de-liberated':constrained, rather than altogether at liberty toresolve the issues in just any old way.1Unconstrained democracy might proceed in any manner that happens to takethe majority's fancy. The constraints of deliberative democracy oblige it to proceed by 'the consideration and discussion of thereasons for and against a measure'.2 Deliberative democracy is obliged to respectthe twin constraints of fact and logic; it is obliged by precepts of sound judgment 'toconsider carefully' the issues at hand. Deliberative democracy, secondly, is deliberate: 'studied; not hasty orrash'. To deliberate is, among other things, 'to take time forconsideration'; and 'deliberate' decisions are 'leisurely, slow, nothurried'. Deliberation itself is likewise defined by an 'absence of hurry' or 'leisureliness'. Thus it is a slow andsteady process, the opposite of a 'rush to judgment'.

1Sayeth Hobbes,Leviathan, ch. 6: '... it is called deliberation; because it isa putting an end to the 2On the secondOxford English Dictionary definition of 'deliberation'. All subsequent quotations,

liberty we had of doing, or omitting, according to our appetite, or aversion'.

unless otherwiseattributed, come from the OED definitions of 'deliberate' (v), 'deliberate'(a) or 'deliberation'.

3 Deliberative democracy, thirdly, is intentional in both its inputs and itsoutputs. Deliberating involves 'weighing in the mind' variousconsiderations; and those considerations being weighed speak to people'spurposes and how alternative resolutions might bear upon them. That is the sense in which people's intentions serve asinputs to deliberation. As regards its outputs, deliberation (democraticor otherwise) gives rise to a 'decision' or 'resolution ordetermination'.3 Those determinations in turn inform people's subsequent intentional actsin pursuing of their various purposes. That is the sense in which theoutputs of deliberation are intentional.4 Turning the phrase on its axis, to say that deliberation ought be'democratic' is, first and foremost, to insist that it accord equal respectto all deliberators (or anyway to all points of view represented amongthem). Each ought be accorded a fair hearing. None ought enjoy any privileged position visa-vis any other. What is eventually adopted as thedeliberators' collective point of view for purposes of any subsequent jointaction ought to have emerged as the fair outcome of a fair competitionamong the deliberators' competing points of view. That is the main point that those advocating 'democraticdeliberation' mean to be making, by invoking that phrase.5 But beneath that pointlies another

3To deliberateis 'to consider carefully with a view to decision'; actions donedeliberatelyare 'done

of set purpose'. That is the second sense in which deliberatingis de-liberating: in the words of Hobbes's Leviathan, ch. 6, 'Everydeliberation is ... said to end, when that whereof they deliberate, is either done, or thought impossible;because till then we retain the liberty of doing, or omitting; according toour appetite or aversion'. 4Hobbes, Leviathan, ch. 6, saysdeliberation issues in 'will'. 5See, e.g.: Joshua Cohen, 'Deliberation and democratic legitimacy', The Good Polity, ed. Alan Hamlin and Philip Pettit (Oxford: Blackwell,1989), pp. 17-34; John S. Dryzek, Discursive Democracy (Cambridge: Cambridge University Press, 1990); Seyla Benhabib,'Deliberative rationality and models of democratic legitimacy', Constellations, 1 (#1: April 1994), 26-52 at e.g. p. 31; Amy Gutmann andDennis Thompson, Democracy and Disagreement(Cambridge, Mass.: Harvard University Press, 1996); and, more generally,the papers collected in James Bohman and William Rehg, eds, Deliberative Democracy (Cambridge, Mass.: MIT Press, 1997) and JonElster, ed., Deliberative Democracy (Cambridge: Cambridge University Press, 1998).

4 equally important point, implicit within that other. To saythat we want to deliberate democratically is to say that we want todeliberate collectively to come to some determination that will be recognized asauthoritative among all deliberators taken as a whole. 'Deliberation', asnoted above, is a process that can all occur within a single person's head.'Democratic deliberation', in contrast, seems necessarily to have an interpersonalcomponent to it.

II. Two Aspects of Deliberation

That last observation points to the twofundamentally different aspects of deliberation which arecentral to my concerns here. On the one hand, deliberation has an'internal-reflective' aspect. Insofar as deliberation is a matter of'weighing and judging reasons for and against any given course of action',all that can take place within the head of a single individual. On the other hand, deliberation has an 'external-collective' aspect. Thesort of give-and-take involved in 'weighing reasons for and against' makesdeliberation an essentially argumentative, and hence discursive, notion.Even where it proceeds entirely within a single person's head, such an internaldiscourse is inevitably modelled upon, and in that way parasitic upon, aperson's experiences of interpersonal discussions and debates with oneanother.6 That makes deliberation a fundamentally interpersonalnotion, at root. It is that external-collective aspect of deliberation which advocatesof democratic deliberation ordinarily want to bolster. They want theupshots of deliberation to be regarded as authoritative by parties to the deliberation and to
6Hence thesecond Oxford English Dictionary definition of 'deliberation' as 'theconsideration and

discussion of the reasons for and against a measureby a number of councillors' (my emphasis). See in this connection Aristotle's remarks on'deliberative speaking' addressed to assemblies in book 1, chapters 3-4 ofRhetoric.

5 form the basis of (at least partially) 'collectiveintentions' thereafter.7 For deliberative democrats those outputs are authoritative only because,and only so far as, they have emerged through external-collective processesof democratic deliberation, involving a free and equal interchange amongeveryone who will be affected by them.8 That ideal seems eminently feasible in small-scale societies whereface-toface interactions are the norm.9 In large-scale mass societies, they are not and cannot be. Dahl offers this calculation, by way of a reductio adabsurdum: 'if an association were to make one decision a day, allow tenhours a day for discussion, and permit each member just ten minutes rather extreme assumptions ... then the association could nothave more than sixty members'.10 Thus, the challenge facing deliberative democrats is to find some way ofadapting their deliberative ideals to any remotely large-scale society,where it is simply infeasible to arrange face-to-face discussions acrossthe entire community.11

7Gutmann and Thompson, Democracy & Disagreement, pp 4-5. HenryRichardson, 'Democratic 8See, e.g.: Bernard Manin, 'On legitimacy and politicaldeliberation', Political Theory, 15 (1987),

intentions', Deliberative Democracy, ed.Bohman and Rehg, pp. 349-82.

338-68; Joshua Cohen, 'Deliberation and democratic legitimacy',esp. pp. 21-3 and 'Procedure and substance in deliberative democracy', Democracy & Difference, ed. Seyla Benhabib (Princeton, N.J.: Princeton University Press, 1996),pp. 95-119 at pp. 99-100. 9Classically explicated in Peter Laslett, 'The face to face society', Philosophy, Politics and Society, ed. Peter Laslett, 1st series. Oxford:Blackwell, 1956), pp. 157-84. 10Robert A. Dahl, Afterthe Revolution? (New Haven, Conn.: Yale University Press, 1970), pp. 67-8 concludes fromthat that the ideal of everyone having 'a full and equal opportunity toparticipate in all decisions and in all the processes of influence,persuasion and discussion that bear on every decision... can exist only among a very small number of people', with'several dozen' marking 'the limits of physical possibility'. 11In the famous phrases of James Madison: 'In a democracy the people meetand exercise government in person; in a republic they administer it bytheir representatives and agents. A democracy consequently, must beconfined to a small spot. A republic maybe extended over a large region' (The Federalist, no. 14). Even where everyone can meetface-to-face, assemblies cease being deliberative when they become toolarge, with speech-making replacing conversation and rhetorical appeals replacing reasoned arguments. As Madison (or perhaps itwas Hamilton) writes, 'In all very numerous assemblies, of whatevercharacters composed, passions never fail to wrest the sceptre from reason.Had every Athenian assembly been a Socrates, every Athenian assembly

6 Various responses have been offered, all of which seem to suffer certaindefects. Having surveyed those, I offer another. My proposal is to easethe burdens of deliberative democracy in mass society by shifting much ofthe work of deliberation from the'external-collective' back into the 'internalreflective', back inside thehead of each individual. In defence of my proposal, I recall that suchinternal mental processes play a very major role even in ordinary conversational settings. It is a small step from there to suggest thatempathetic imagining can substantially substitute for interpersonalconversation in the sorts of deliberations which democrats desire acrossmass societies. On that view, the challenge confronting deliberative democrats in mass society lies less in makingeveryone else 'conversationally present' than in making them 'imaginatively present' in the minds of deliberators.12 This alternative model has the further effectof prioritizing what might otherwise seem peripheral to modern democratic theory. Nodoubt democrats always regard it as a presumptively good thing todemocratize all corners of society, the arts along with everything else.But seeing empathetic imaginings as central to the deliberative processes of mass democracies sensitizes us toconditions surrounding the production and distribution of crucial aids tothose imaginings, conspicuously among them the literary, visual andperforming arts. Ensuring the broad representativeness of those representations is, on the model of democraticdeliberation I am propounding, of capital importance.
would still have been a mob' (TheFederalist, no. 55). Cf. Cicero, De Officiis, II.48 and Madison, The Federalist, nos. 62, 63. Among those who see size as the mainchallenge to deliberative democracy are: Robert A. Dahl and Edward R.Tufte, Size and Democracy (Stanford, Calif.: StanfordUniversity Press, 1973); Jane J. Mansbridge, Beyond Adversary Democracy (New York: BasicBooks), chs. 19-20; James S. Fishkin, The Voice of thePeople: Public Opinion and Democracy (New Haven, Conn.: Yale UniversityPress, 1995), ch. 2; Benjamin I. Page,Who Deliberates? Mass Media in Modern Democracy(Chicago: University of Chicago Press, 1996), ch. 1. 12After the fashion of Benedict Anderson's Imagined Communities (London: Verso, 1983). On imagining in general, seeGilbert Ryle, The Concept of Mind (London: Hutchinson, 1949), ch.8.

7 III. Problems with Previous Solutions

Previous solutions to the problems posed fordeliberative democracy by mass society seem to rely upon one of four basicstrategies. The first two work by limiting the number of people with whom you have to deliberate, the second two by limiting the levelof inputs from others with which you have to deal. Each of thesestrategies seems likely to fail, for one reason or another.

A. Seriality: Disjointed Deliberation

If we are too numerous to deliberate together all atthe same time, then one solution might be found in Aristotle's suggestionthat we 'deliberate, not all in one body, but by turns'.13 Here is one way. Break us down into groups sufficiently small to allowgenuine deliberation within each of them; and then let the upshots of thosedeliberations serve as inputs to subsequent deliberations among othergroups, similarly constituted. That is nothing more than a stylized accountof how the 'common law' has been supposedly being discovered by Englishjuries since they were formally instituted by Henry II in the twelfthcentury.14 True, modern juries have us 'deliberating with one another' only a dozenat a time (even the largest ancient ones had Athenians deliberating withone another only five hundred or, at most, a thousand at a time15). But now suppose that all potential jurors serve on ajury
13Aristotle, Politics,1298a13. 14Harold J.Berman, Law and Revolution: The Formation of the Western LegalTradition (Cambridge,

Mass.: Harvard University Press, 1983), pp. 448-9. Juriesenjoy a privileged place in contemporary discussions of deliberativedemocracy: apart from the increasingly quaint New England town meeting,juries are the sole institution which 'regularly calls upon ordinary citizens to engage each other in aface-to-face process of debate' (Jeffrey Abramson, We, the Jury (NewYork, Basic, 1994), p. 8). 15Aristotle, Constitutionof Athens, 68.

8 several times over the course of their lives16; andfurther suppose that no two jurors ever serve together more thanonce.17 Then each individual would be deliberating with one group on oneoccasion, and another on another. Through several iterations of thisprocess of pooling one's wisdom, first with one group and then with thenext, we could come to some 'common' view (about the content of the 'common law', or whatever) across the community as awhole.18 Let us therefore call this a model of 'serial' or 'disjointeddeliberation'. A variation on that model appeals to certain sorts of post-modernists. The problem, asthey see it, is an increasingly fractured social world. One solution, theysometimes suggest, might be a 'directly-deliberative polyarchy' involving'a plurality of modes of association'.19 The trick, of course, lies in how all those separately deliberating bodies'judgments can be articulated with one with another. There are many ways,not all of them particularly democratic. The deliberations of localEnglish assizes, for example, were made into a unified common law across the realm by the same smallset of judges (twenty five in all) travelling up and down the

16As originally they did, when the same handful of local notables assembledin periodic English

assizes. What matters of course is the ratio ofpotential jurors to the number of trials for which they are needed: anincrease in the number of potential jurors can be counterbalanced by an increase the number of trials (which isalso happening, of course, in our increasingly litigious societies). 17As was clearlynot the case in the original English assizes, where the same local notablesmet together time and again, never crossing county boundaries. 18For a back-of-the-envelope calculation, let us say that each person poolshis opinions with (N-1)i other people, where N is the sizeof the deliberating unit and i is the number of different groups with which each person deliberates.Supposing each of us deliberated with a group of 11 others on our every tenth birthday, from the ages of 20 to 70, then our opinion wouldhave thereby been pooled (directly or indirectly) with those of116 or 1,771,561 other people. 19Joshua Cohen and Charles Sabel, 'Directly-deliberativepolyarchy', European Law Journal 3 (#4: Dec 1997): 313-42. See similarly: Joshua Cohen and Joel Rogers etal., Associations and Democracy, ed. E. O. Wright (London: Verso,1995); Paul Hirst, Associative Democracy (Oxford: Polity, 1994); Seyla Benhabib, 'Deliberative rationality andmodels of democratic legitimacy', p. 35; and Iris Marion Young,'Communication and the other: beyond deliberative democracy', Intersecting Voices (Princeton, N.J.: Princeton University Press, 1997), pp. 60-74.

9 countrypresiding over all proceedings. Sir Matthew Hale proudly proclaims thatthat: keeps both the Rule and Administration of the law of the kingdom uniform;for those men are employed as justices, who as they have had a commoneducation in the study of law, so they daily in term-time converse andconsult with one another; acquaint one another with their judgements, sit near one another in Westminster Hall, whereby their judgements arenecessarily communicated to one another, and by this means their judgementsand their administrations of common justice carry a constancy, congruityand uniformity one to another, wherebyboth the laws and the administrations thereof are preserved from theconfusion and disparity that would unavoidably ensue, if the administrationwas by several uncommunicating hands, or by provincialestablishments.20

Deliberative democrats, however, are unlikely to share Hale's enthusiasm for that system. From theirperspective, whatever directly deliberative gains were secured byintroducing juries in the first place would be largely nullified byentrusting such a small and closed elite with the task ofblending the products of all those lower-level deliberations into a singlenationwide common law. The upper English judiciary, however, constitutes onlythe most dramatic instance of a problem that plagues models of disjointeddeliberation quite generally. Many of them seem to involve aggregating the inputs of highly democraticground-level groups through a hierarchy which itself has substantially lesssecure directly-democratic deliberative credentials. Consider, forexample, Habermas' scheme for an 'oppositional public sphere' to serve as the source ofdeliberatively-democratic inputs, which are then fed into and coordinatedthrough the ordinary political sphere which (Habermas would be the

20Sir MatthewHale, The History of the Common Law of England, 2nd edn, ed. CharlesM. Gray

(Chicago: University of Chicago Press, 1971; originally published1716), p. 252; quoted in A. W. B. Simpson, 'The common law and legal theory',Oxford Essays in Jurisprudence, 2nd series, ed. A. W. B. Simpson(Oxford: Clarendon Press, 1973), pp. 77-99 at p. 96. See also Berman,Law in Revolution, p. 449.

10 first toadmit) is itself far from being directly deliberatively democratic.21 Or consider, again, Cohen and Sabel's proposal for the outputs of'directly deliberatively democratic polyarchies' at the local level to befed into a 'peaklevel' meta-deliberation among all those groups: onceagain, however equal the groups might be within that meta-deliberation, the incorporation of that extra layer ofdeliberation itself makes that scheme less 'directly deliberative', andhence less democratic.22 Much thesame might be said about the various other mediating institutions whichhave from time to time been proposed as key agents in the deliberativeprocess.23 A sociological twist on the jury strategy of disjointed deliberationdescribed above, however, suggests one way in which the inputs of aplurality of groups might indeed be blended together in a fashion thatseems both directly and deliberatively democratic. Suppose each of us will be a member of many different groups at once;and further suppose that each of us overlaps any given other in only asmall fraction of our group memberships.24 Then there might be a 'web of group affiliations' which links(indirectly, perhaps very indirectly) everyone with everyone else in adialogue which will (eventually) effectively straddle the entirecommunity.25 That way of linking all the groups
21Jrgen Habermas, Between Facts andNorms, trans. William Rehg (Oxford: Polity, 1996;

originally published 1992),ch. 8. See similarly John S. Dryzek, 'Political inclusion and the dynamicsof democratization', American Political Science Review, 90 (1996),475-87. 22Cohen and Sabel, 'Directly-deliberative polyarchy', p. 326. 23These rangefrom political parties to legislative subcommittees to high courts. See,e.g., Cohen, 'Deliberation and democratic legitimacy', pp. 31-2; Joseph M.Bessette, The Mild Voice of Reason: Deliberative Democracy and American NationalGovernment (Chicago : University of Chicago Press, 1994), esp. ch. 6; andJohn Uhr, Deliberative Democracy in Australia (Cambridge: Cambridge University Press, 1998), esp. ch 4. 24Something like this is suggested, e.g., byIris Marion Young's talk of 'groups ... as overlapping, ... constituted inrelation to one another and thus ... shifting their attributes and needs in accordince with what relations are salient';'Together in difference: transforming the logic of group politicalconflict', The Rights of Cultural Minorities, ed. Will Kymlicka (Oxford: Oxford University Press 1995), pp. 155-77 atat p. 157. 25Georg Simmel, Conflictand the Web of Group Affiliations , trans. Kurt H. Wolff and ReinhardBendix (Glencoe, Ill.: Free Press, 1955), esp. pp. 125-95.

11 does seem genuinely to hold thepromise of being both deliberatively and directly democratic. That is what I take to be the bestconceivable model of 'disjointed deliberation'. But even that model of'disjointed deliberation' depends crucially upon certain presuppositions which are likely not to be met in thereal world. Specifically, it presupposes that everyone is a member of some(indeed, several) groups, each of which approximates the deliberativeideal. In the real world, altogethertoo many people are socially excluded (participating in no properlydeliberative groups within the community) or socially segregated(participating in only the same sets of deliberative groups). Insofar aseither is the case, there will to that extent beno 'serial deliberation' including, however indirectly, all the members ofthe community. The weakness of those presuppositions marks the limits ofthe model of disjointed, serial deliberation in the real world.

B. Substitution: Ersatz Deliberation

'Disjointed deliberation' substitutes deliberationwithin partial, overlapping groups for deliberation across the entirecommunity. Models of 'ersatz deliberation', in contrast, substitutedeliberation within a subset of the community for deliberation across the whole of the community. How the subset is identified and how the substitution isjustified are interconnected issues. The subset is supposed to berepresentative typical, 'a

In a stylized calculation akin to thatperformed for juries above, if each ofus were a member of just five groups containing twenty others each, andnone of us overlapped any other more than once, then on these assumptionsour judgment would be merged (directly or indirectly) with 205or 3,200,000 others.

12 fair sample', 'a microcosm'26 of the larger group. Substituting its judgment for that of the larger isjustified, in turn, on the grounds that the considered views reachedthrough deliberation within that smaller group will be representative (anaccurate reflection) of the views that would have been reached had similar processes been feasible within thelarger group.27 The clearest example of this sort of 'ersatz deliberation' the substitution of deliberation within a smaller group for that within anunwieldily large one is of course representativedemocracy.28 Legislatures are continually styled as 'deliberative assemblies', incontrast to 'popular' ones (and even within the legislature, the lessnumerous house is standardly styled the 'deliberative' chamber29). But recent innovations like 'citizens'juries' and 'deliberative polling' are other instances of the same broadclass of model.30 In the realm of high theory, models structurally similar to these standbehind the 'thought experiments' of Rawls and Habermas. In both thoseauthors' earlier writings, anyway, the issue is not what any actual peopleactually decide. Rather, the issueis what propositions would

26John Stuart Mill, Considerations on Representative Government (1861), ch. 5; inRichard Wollheim,

John Stuart Mill, Three Essays (Oxford: ClarendonPress, 1975), pp. 142-423 at p. 228. In Lord Boothby's blunter formulation, 'Ideally,the House of Commons should be a social microcosm of the nation. The nation includes a great many people who arerather stupid, and so should the house'; quoted in A. H. Birch, 'Thenature and functions of representation', The Study of Politics: ACollection of Inaugural Lectures, ed. Preston King (London: Frank Cass, 1977), pp. 265-78 at p. 268. 27Thus, for example, in their deliberations behind the closed doors of thePhiladelphia Convention the Founding Fathers self-consciously couched theirarguments in terms of what 'ought to occur to a people deliberating on a Government for themselves, ... in a temperate moment,and with the experience of other nations before them'; James Madison, Notes of Debates in the Federal Convention of 1787 (New York: Norton, 1966; originally published 1840), entry for 26 June1787, pp. 193-4. 28Hanna F.Pitkin, The Concept of Representation (Berkeley: University ofCalifornia Press, 1967). Bernard Manin, Principles of RepresentativeGovernment (Cambridge: Cambridge University Press, 1997), esp. ch. 6. 29Madison, TheFederalist, nos. 62, 63. 30James S.Fishkin, Democracy and Deliberation: New Directions for DemocraticReform (New Haven, Conn.: Yale University Press, 1991) and The Voiceof the People. Anna Coote and Jo Lenaghan, Citizens' Juries: Theory into Practice(London: Institute for Public Policy Research, 1997).

13 hypothetically command universal assent inspecific choice situations, idealized in certain respects. The early Rawlsand Habermas would have us substitute the hypothetical deliberations of individuals in those idealized choicesituations for the actual outcomes of actual political processes, flawed asthey are by ignorance, self-interest and so on.31 All these models of ersatz deliberation involve, at root, substitutinga subset for the whole and letting the subset deliberate on behalf of thewhole. The generic problem with all of those schemes lies in ensuring thecontinuing representativeness of the subset, once the deliberation gets underway. Ofcourse people change their minds in the course of the deliberation (itwould hardly be a genuine deliberation at all if they did not, from time totime32). The question is whether people who started out being representative ofthe wider community, in all the ways we can measure, are alsorepresentative of that wider community in the ways in which they change over the course of the deliberation. On the face of it, that seems unlikely. From everyday life we knowthat different conversations with different participants (or the sameparticipants interjecting at different points) proceed in radicallydifferent directions.33 Given

31John Rawls, A Theory ofJustice (Cambridge, Mass.: Harvard University Press, 1971).

JrgenHabermas, The Theory of Communicative Action, trans. Thomas McCarthy(Boston: Beacon Press, 1984).Later iterations of these theories, both from the principals and theirfollowers, succeed in making the deliberations more 'real' but in ways thatrun into other of the problems here discussed. How this happens inHabermas' Between Facts and Norms have been discussed already; and similar difficulties arise with theproblem-centred approach suggested by William Rehg, 'Intractable conflictsand moral objectivity: a dialogical, problembased approach', Inquiry, 42 (#2, June 1999), 229-58. Rawls, Political Liberalism (New York: Columbia University Press, 1993) purports to make thederivation of principles of justice 'political, not metaphysical': but thediscussions are still imaginary, not real (the issue is what reasonablepeople could assent to, not what they do); and even that imaginary deliberation is 'blinkered' (in terms that willbe introduced below) insofar as certain positions are ruled out of courtas being 'unreasonable'. See similarly T. M. Scanlon, What We Owe toEach Other (Cambridge, Mass.: Harvard University Press, 1998). 32Elster, Deliberative Democracy, esp. pp. 8-9. 33Thetheoretical importance of which, in other connections, is teased out byJames Tully, 'The agonic freedom of citizens', Economy & Society, 28(#2: May 1999), 101-22.

14 the path dependency of conversational dynamics, and the sheercreativity of conversing agents, it beggars belief that any one group wouldcome to exactly the same conclusions by exactly the same route as anyother.34 (Lawyers sayit is a 'well known secret' that 'no two juries and no two judges arealike'.35) Yet that is what strong advocates of ersatzdeliberation must be claiming, in insisting that deliberations within a representative subset will genuinelymirror, and can therefore substitute for, deliberations across the wholecommunity.36

34Of course we will never know whether the proposition that the deliberationsof the subset end

up the same place as would deliberations of the wholewithout undertaking the latter deliberations for which the former weresupposed to substitute. But short of that there are some more modest empirical experiments which we mightundertake to test the proposition. We might, for example, provide exactlythe same briefing books and exactly the same videotaped testimony toseveral deliberating groups, to see if they reach the same conclusions given the same evidence. Fishkin, Voice of the People, p. 220, reports the results ofa 'natural experiment' of that sort, induced by the requirement that localpublic utilities in Texas consult the public (which in three cases reported there was done through deliberative polling) beforeinstituting new policies. In all three cases, the pre- topost-deliberative shift was in the same direction, at least on three verybroad questions reported there. But the absolute numbers diverged wildly: in one place half the respondants, post-deliberation, thought that 'investing in conservation' was the 'option to pursuefirst to provide additional electrical power to the service territory',whereas in another under a third thought so; in one place, over a third still thought post-deliberation 'renewableenergy' should be the first option, in another place under half that manythought so. There was rather more uniformity in whether people were'willing to pay at least $1 more on their monthly bill for renewable energy': but whereas the pre- topost-deliberative shift in favour of doing so was merely 58 to 81 percentin one place, it was 56 to 90 percent in another, a shift of 23 percent inthe first instance but 34 percent in the second. These really are not the sorts that would encourage us toconclude that these deliberating groups ought be regarded asinterchangeable. Nor do they inspire confidence, in consequence, in thegeneral strategy of 'ersatz deliberation', allowing smaller deliberative groups as microcosms capable of literally'substituting' for deliberation across the whole community. 35Harry Kalven,Jr., and Hans Zeisel, The American Jury (Chicago: University ofChicago Press, 1966), p. 474. Their mock-jury experiments revealed certain general statistical regularities: but theregularities were purely statistical; in no sense did each jury do exactlythe same thing. 36At most, they might be taken as 'recommendations' to be fed back into thosebroader community-wide deliberations, as Fishkin puts it in Voice ofthe People, p. 162.

15 C. Restricting Inputs: Emaciated Deliberation

Whereas the first two models cope with the problemof mass society by reducing the number of people deliberatingtogether, the second class of models cope with the problem by reducing how much they communicate to one another. The first variation on this model is one of 'emaciateddeliberation', which facilitates mass deliberation by reducing the densityof the signals and hence the deliberative load each participant has tobear. Legislative assemblies, for one familiar example, havelong been accustomed to streamlining their proceedings by imposing limitson the length or number of speeches.37 One effect (among others) is to limit how much speakers can say, andhence to restrict how much input deliberators have to take into account.Rules of 'germaneness', of course, serve even more directly to filter the quantity (as well, obviously, as the quality) of inputs intolegislative deliberations.38 Another version of this process is 'mediated deliberation', wherethere is some intermediary who filters the messages that get passed alongto others within the larger community.39 In internationalnegotiations, intermediaries facilitate agreement simply by censoring themessage (omitting gratuitous insults and groundless threats and soon).40 In modern mass society, much the same sort
37Limiting the length of interventions is the more modernway, limiting the number of them the

older. In British parliamentarypractice, traditionally 'none may speak more than once to the matter';Thomas Jefferson, Parliamentary Pocket-Book; reprinted in Jefferson's ParliamentaryWritings, ed. Wilbur Samuel Howell, 2nd series (Princeton, N.J.:Princeton University Press, 1988), pp. 47-162 at sec. 180, p. 89. 38Jefferson's Parliamentary Pocket-Book, sec. 180, p. 89-90, goes on to say, 'The Speaker collecting the sense ofthe house upon the debate, is to reduce the same into a question, which heis to propound, to the end the house in their debate afterward may be keptto the matter of the question, if the same be approved by the house to contain the substance ofthe former debate'. 39Page, Who Deliberates? 40Oran R. Young, TheIntermediaries: Third Parties in International Crises (Princeton, N.J.:Princeton University Press, 1967).

16 of 'mediated deliberation' occurs through the agency of the mass media. In the 'democracy ofsoundbites', there are very strict limits on how much information anyonecan impart to (or impose upon) everyone else. There are obvious problems, however, with all such strategies forfacilitating mass deliberation by 'restricting inputs' into them. Unlesswe have some reason for supposing we are screening out only inputs whichare irrelevant or superfluous41,restricting inputs leaves us deliberating more-or-less in ignorance. Ourcognitive capacities, which rely upon informational inputs, are more-or-lessundernourished. (Hence the term, 'emaciated deliberation'.) In thelimiting case of a 'democracy of soundbites', we are deliberating on thebasis of so little as to make it hardly acase of deliberation of seriously reflective 'weighing and judgingreasons' at all.

D. Selective Uptake: Blinkered Deliberation

The most famous deliberative democrat of the modernera, perhaps, is Jrgen Habermas. Much recent work in this area borrowsand builds on his notions of 'civil society' and the 'publicsphere'.42 This is not theplace to elaborate his larger theories. For the narrower purposespresently at hand, let us focus upon what exactly is supposed to happen inthe 'public sphere'. Ideally, of course, we are supposed to 'engage' with one another. Insome of the early institutions out of which the modern public sphere grew such as
41As arguably we do in the legislative case. 42Habermas, The Structural Transformation of the Public Sphere, trans. Thomas Burger and

Frederick Lawrence (Oxford: Polity, 1989;originally published 1962), esp. pp. 31-43; 'The public sphere' (trans. S.and F. Lennox), New German Critique, 3 (1964), 49-55; and BetweenFacts and Norms, esp. chs 7-8. Note the way these themes have been taken up in, e.g.,John Rawls, 'The idea of public reason revisited', University ofChicago Law Review, 94 (1997): 765-807, and Charles Taylor, 'Modernityand the rise of the public sphere', Tanner Lectures on Human Values, 14 (1993), 203-60.

17 the coffee-houses of eighteenth century London, forexample people may well have done just that. (They engaged directly, but oncontemporaneous accounts not particularly deeply.43) Most of the institutions of thecontemporary public sphere, however, are much more nearly like Habermas'other great paragon of the early public sphere, the broadsheet newspaper. 'When Addison and Steele published the first issue of the Tattlerin 1709', Habermas says, 'the coffee-houses were already so numerous andthe circles of their frequenters already so wide, that contact among thesethousandfold circles couldonly be maintained through a journal'.44 Participants in the public sphere were no longer all engaging directlywith one another. Their engagements were mediated through the broadsheet,in ways I have already described and with the problems I have alreadysketched. Here, however, I want to draw attention to yetanother problem, which is just this: Contributors to broadsheets as, indeed, those holding forth in coffeehouses are not so much 'talking to one another' as they are 'posting notices for all to read'. Others might (or might not) take note of them, and reply. Butinsofar as they reply in similarly public fashions, they too areessentially just posting other notices for all to note (or not), in turn.
43Consider the descriptions in William Hazlitt's essay 'On coffee-housepoliticians', Table Talk, or

Original Essays (New York: ChelseaHouse, 1983; originally published 1869), pp. 261-83. Coffee-house politicians,Hazlitt says, 'are like an oyster at the ebb of the tide, gaping for freshtidings' (p. 263). Among them, 'The Evening Paper is impatiently expected andcalled for at a certain critical minute: the news of the morning becomesstale and vapid by the dinnerhour. .... It is strange that people shouldtake so much interest at one timewhat they so soon forget: the truth is, they feel no interest init at any time, but it does for something to talk about. Their ideas are served up to them, like their billof fare, for the day' (p. 262). In coffeehouses, furthermore, 'People donot seem to talk for the sake of expressing their opinions, but to maintainan opinion for the sake of talking. .... It is not conversation, but rehearsing a part' (pp. 268-9).Finally, the coffee-house politician 'goes around for a meaning, and thesense waits for him. Men of education and men of the world order thismatter better. They know what they have to say on a subject, and come to the point at once. Your coffeehousepolitician balances between what he heard last and what he shall say next;and not seeing his way clearly, puts you off with circumstantial phrases,and tries to gain time for fearof making a false step' (p. 269). 44Habermas, Structural Transformation, p. 42.

18 What we find in the public sphere, in short, isnot so much 'public deliberation' as 'deliberation in public'.Beijing's 'Wall of Democracy' was democracy of a sort. And democracy ofthat sort the free broadcasting of opinions might be one importantprecondition of any genuinely deliberative democracy. But just posting notices, just shouting out opinions from asoapbox in Hyde Park, does not in and of itself constitute communication,much less fullblown deliberative democracy.45 There must also be uptake and engagement other people musthear or read, internalise and respond for that public-sphereactivity to count as remotely deliberative.46 Furthermore, for that public-sphere to count as particularly democratic,it must be the case that most people are actively engaged in this sort ofgive-and-take with most other people. Theorists of the public sphere, in short, solve the problem of how todeliberate democratically in mass societies by compromising the conditionsthat make the processes deliberative. In guaranteeing the free and equalexpression of opinions in the public sphere, they guarantee everyone a voice but no one a hearing.

45British parliamentary practice prohibitsreading out of written speeches for just that reason:

'When oratorsconfine themselves to reading out what they have written in the silence of their study, they no longer discuss, they amplify. They donot listen, since what they hear must not in any way alter what they aregoing to say. They wait until the speaker whose place they must take hasconcluded. They do not examine theopinion he defends, they count the time he is taking and which they regardas a delay. In this way there is no discussion... Everyone sets asidewhatever he has not anticipated, all that might disrupt a case alreadycompleted in advance. Speakers follow one another without meeting; if they refute one another it is simply bychance. They are like two armies, marching in opposite directions, onenext to the other, barely catching a glimpse of one another, avoiding evenlooking at one another for fear of deviating from a route which has already been irrevocably traced out'.Benjamin Constant, 'Principles of politics applicable to all representativegovernments', Political Writings, trans & ed. Biancamaria Fontana (Cambridge: Cambridge University Press,1988; originally published 1815), ch. 7, p. 222. 46On uptake see J. G. A. Pocock, 'Verbalizing apolitical act: toward a politics of speech', Political Theory, 1(1973): 27-45.

19 IV. Deliberation Within

All of those previous proposals for makingdeliberation work in the context of mass society focus on the'external-collective' side of deliberation. They all suppose that the key to makingdemocracy deliberative is making everyone 'communicatively present', insome sense or another. But in any large society, it is impossible to dothat literally, and it seems unsatisfactory to do that through any of the various other devices that have beensuggested so far. My proposal is to go back to the beginning, to see if wecannot make the 'internal-reflective' aspect of deliberation do more of thework for us. Let us begin by recalling how very much of what goes on in a genuine face-to-face conversationis actually contained inside the head of each of the participants, anyway.Even if there is a good argument to show that language is essentiallyshared rather than private, stillit is the case that most of the work in interpreting the utterances ofothers decoding the literal meaning, and enriching that literalmeaning pragmatically in light of contextual information is donewithin the hearer's own head. Thus, for example, when trying to understand what others are saying, we start by assuming thatthey are trying to talk sense and, at least as a first approximation, weassume that they mean by their utterances roughly what we ourselves wouldhave meant by them.47 We are prepared to treat provisionally as true, for the purposes of anygiven conversation, that set of propositions which

47J. L. Austin, 'Other minds', Philosophical Papers, ed. J. O. Urmson and G. J. Warnock, 3rd edn

(Oxford: Oxford UniversityPress, 1979), pp. 76-116 at p. 115. Donald Davidson, Inquiries intoTruth and Interpretation (Oxford: Clarendon Press, 1984).

20 seem to constitute themost coherent way of construing the background assumptions underlying theassertions our interlocutor is making.48 In ordinary conversation, people do not tediously elaborate completesyllogisms. (Nobody listens, if they do.) Instead, peoplecharacteristically talk more or less 'loosely'.49 They make more-or-less cryptic allusions to more full-blown arguments.What is involved in your 'catching the other's drift' in ordinaryconversation is completing the other's syllogism in your own mind, workingout the various 'implicatures' containedwithin their conversation.50 What exactly is going on, when we try to make sense of 'other minds',is a large and contentious issue. Some say that in trying to understandothers' behaviour we employ a 'folk psychology'.51 We attribute to others the same sort of psychology of beliefs and desireswhich, upon introspection, we find that we ourselves have. We assume thatthey will act on their peculiar beliefs and desires in standard sorts ofways, under standard sorts of provocations, just as we ourselves (and, from our observation, most others)do. Maybe what we do in trying to explain, predict or understand their behaviour is simply conduct a mental simulationof the mental processes that we assume are going on in their heads insideour own heads. Or maybe what we do is deploy some rudimentarytheories.

48David Lewis, 'Scorekeeping in a language game', Journal of Philosophical Logic, 8 (1979), 339-59. 49Dan Sperber and Deidre Wilson, 'Loose talk', Proceedings of the Aristotelian Society, 86 (1986), 50That is how'conversational implicature' works: G. Paul Grice, 'Logic andconversation', The

153-71.

Logic of Grammar, ed. Donald Davidson and Gilbert Harman (Encino, Calif: DickensonPublishing Co., 1975), pp. 64-75and Studies in the Way of Words (Cambridge, Mass.: Harvard University Press, 1989), esp. pp. 22-40,138-44, 269-82. 51Frank Jackson and PhilipPettit, 'In defence of folk psychology', Philosophical Studies, 57(1990), 730. Philip Pettit, The Common Mind, 2nd ed (New York: Oxford University Press, 1996), esp. chs 1-2, 4 andpostscrip. David Braddon-Mitchell and Frank Jackson, ThePhilosophy of Mind and Cognition (Cambridge, Mass.: Blackwell, 1996).

21 All that touches upon deep issues in the philosophy of language andthe philosophy of mind which for present purposes I want simply togloss.52 What is centrally at issue here is simply that, in their ordinarydealings, people make sense of one another, their utterances and theiractions,by mentally 'putting themselves in the other's place' in some sense oranother. 'Simulation theorists' envisage people 'understanding other mindsfrom the inside' in a lightly theorized way.53 'Theory theorists', as the name implies, envisage them doing so in a muchmore theoretically-laden way.54 But either way, agreat deal of the work involved in making sense of what others are saying,in ordinary discourse, has necessarily to go on in the hearer's ownhead. Discourse theorists know perfectly well thatthis is how ordinary conversation proceeds. Indeed, they make much of thefact. Habermas, for example, describes discourse ethics as 'rest[ing] on... a jointprocess of "ideal role taking"' in which 'everyone is requiredto take the perspective of everyone else, and thus project herself into the understandings of self andworld of all others'.55 That is howhe hopes to get the 'intersubjectivity' that is so crucial to his project.
52On related disputes in cognitive science, see: Gregory Currie, Meetingof Minds: Thought,

Imagination & Perception (Oxford: Oxford UniversityPress, 1999); Christopher Peacocke, ed., Simulation and the Unity ofConsciousness: Current Issues in the Philosophy of Mind, Proceedings of the British Academy # 83 (Oxford: Oxford UniversityPress for the British Academy, 1994), esp. Martin Davies, 'The mentalsimulation debate', pp. 99-127; Martin Davies and Tony Stone, 'Folkpsychology and mental simulation', Current Issues in the Philosophy of Mind, ed. Anthony O'Hear (Cambridge:Cambridge University Press, for the Royal Institute of Philosophy, 1998),pp. 53-82. 53Jane Heal, 'Understanding other mids from theinside', Current Issues in the Philosophy of Mind, ed. O'Hear, pp. 83-100. 'Lightly theorized', but still nonethelesstheorized: we need some grounds for supposing they are like us in relevantrespects, for example. 54Ned Block, ed., Readings in the Philosophy of Psychology (Cambridge, Mass.: Harvard University Press, 1980). 55To reproduce the quotation in full: 'Discourse ethics rests on theintuition that the application of the principle of universalization,properly understood, calls for a joint process of "ideal role taking". It interprets this idea of G. H. Mead in terms of a pragmatictheory of argumentation. Under the pragmatic presupposition of aninclusive and noncoercive rational discourse among free and equalparticipants, everyone is required to take the perspective of everyone else, and thus project herself into the understandingsof self and world of all others'. See Jrgen Habermas, 'Reconciliationthrough the public use of reason: remarks on John Rawls's PoliticalLiberalism', Journal of Philosophy, 92 (March 1995), 109-31 at p. 117.

22 It is precisely that process of imagining yourself in the place of theother which discourse theorists hope discourse will set in motion. Butnote well: The process which discourse theorists valorize is this'internal-reflective' one. The 'external-collective' process of discourseand debate is, even for discourse theorists, merely a means of setting this other more 'internal-reflective' processin motion.56 Having to answer to another in person might well be one good way ofgetting that process going. But it is hardly the only. Sometimes'answering to oneself' can suffice instead. Suppose our imagination has been fired by some film or fiction;we have been led by those artifices to imagine vividly what it would belike to be them, or to be in that situation; we ask ourselveswhat we would say, then. How exactly all that works is the subject of much debate in thephilosophy of mind and cognitive science, literary theory and artcriticism.57 Here again, however, the fact that some such process does seem to be at work seems incontrovertible enough, judging simply from ourown everyday experience as readers, viewers and listeners.58 One particularly striking example is the way in which 'slavenarratives' autobiographical accounts by freed slaves, vividlyevoking their experiences in

56As Iris Marion Young aptly remarks, a propos Habermas, it is not just amatter of intellectually

registering the perspective of the others butrather of 'imaginatively' projecting oneself into their position;'Asymmetrical reciprocity: on moral respect, wonder and enlarged thought', Intersecting Voices, pp. 38-59 at p.39. 57See, e.g.: Richard Wollheim, The Thread of Life (Cambridge:Cambridge University Press, 1984), esp. chs 3-4 ; Gregory Currie, Imageand Mind: Film, Philosophy and Cognitive Science (Cambridge: CambridgeUniversity Press, 1995); and Elaine Scarry, 'On vivacity: the difference between daydreaming andimagining-under-authorial-instruction', Representations, # 52 (Fall1995), 1-26. 58It may well be that one's own 'sense of oneself' is similarly constructedout of some such internal narrative. See, e.g., Charles Taylor, Sources of the Self (Cambridge: Cambridge University Press, 1989), esp.ch. 2, and Alasdair MacIntyre, After Virtue (Notre Dame, Ind.: University of Notre Dame Press,1981).

23 bondage served the abolitionistcause.59 But some such 'expanding of people's sensibilities' occurs in all goodwriting.60 Literary theorists regard it as something of a commonplace that'historical and social events as mirrored in the plots of Stendhal, Dickensor Tolstoy had a realness, anauthenticity deeper than that conveyed by the journalist or professionalhistorian. .... The art of Balzac is a summa mundi, an inventory ofcontemporaneous life. A man can learn half a dozen professions by readingZola'.61 It is not just that fiction (and art more generally) might, and often does,contain allusions to social, economic, political and historical facts, andin that way might serve certain didactic purposes. The larger point isthat those lessons come packed with more emotional punch andengage our imagination more effectively than do historical narratives orreflective essays of a less stylized sort. 'Artists', John Dewey says,'have always been the real purveyors of news, for it is not the outwardhappening in itself which is new, but the kindling by it of emotion, perception and appreciation. ....Democracy', he continues, 'will have its consummation when free socialinquiry is indissolubly wedded to the art of full and movingcommunication'.62 That is not just to say that novelists are more evocative writers thanhistorians or essayists (true though that may be, too). Rather, they fixtheir focus on the particular one person or one action or oneperiod and they

59Kimberly K. Smith,'Storytelling, sympathy and moral judgment in American abolitionism', The 60This, and the political implications flowing from it, form recurringthemes in the writings of

Journal of Political Philosophy, 6 (1998), 356-77.

Martha Nussbaum. Cf. Love's Knowledge:Essays on Philosophy and Literature (New York : Oxford University Press,1990); Poetic Justice: The Literary Imagination and Public Life (Boston: BeaconPress, 1995); and Cultivating Humanity: A ClassicalDefense of Reform in Liberal Education (Cambridge, Mass : HarvardUniversity Press, 1997). 61George Steiner, 'Liberature and post-history' (1965), Language andSilence (London: Faber & Faber, 1967), pp. 413-24 at p. 420. 62John Dewey, The Publicand Its Problems (Chicago: Swallow Press, 1954; originally published1927), ch. 5, p. 184.

24 introduce generalities as anecdotes, episodes viewed from thatparticular one's perspective.63 That vivid evocation of the particular, in turn, has importantconsequences for the uptake of works of art. Inevitably, we find itrelatively easy to project ourselves imaginatively into the place of somespecific (fictitious but grounded) other. It is necessarily harder to project ourselvesimaginatively into the inevitably underdescribed sorts of amorphous sets ofabstract others which are the stock in trade of historians and socialscientists.64 That fiction 'takes us out of ourselves' in this way is intrinsicrather than incidental to the enterprise: The very form of the novel arises from and embraces conflictsof character, values , interests, circumstances and classes. And the formthat only seems to work well, as Sartre pointed out in his What isLiterature?, when the author has at least an implicit commitment to freedom and canempathize with and portray plausibly social diversity. Where, he asked, isthere a great totalitarian novel? That was not just a contradiction interms but a psychological impossibility.65

Thus, as George Steiner suggests, To read well is to take great risks. It is to make vulnerableour identity, our self-possession .... [W]hen we take in hand a major workof fiction or philosophy, of imagination or doctrine, ... [i]t may come to possess us so completely thatwe go, for a spell, in fear of ourselves...66

63Aristotle, Poetics,1459a17-1459b8. 64Conduct theexperiment for yourself. Is it not ever so much easier to imagineyourself Jean

Valjean, given what all Hugo has told us about him, than it is to imagine yourself a generic 'prisonerof the Bastille' on the basis of what historians have told us about thatplace and its denizens? Intellectually, generalizations may be easier,both to convey and to grasp; but emotionally and imaginatively, it is easier to evoke more full-describedparticulars than generalities which abstract from the detail that makethose particulars more easily imagined. 65Bernard Crick, Essays on Politics and Literature(Edinburgh: Edinburgh University Press, 1989), p. 17. 66Steiner,'Humane literacy' (1963), Language and Silence, pp. 21-9 at p. 29.

25 That rhetoric may seem overblown, but the less prosaic truth which isnonetheless telling is simply this: the 'unique value of fiction' lies inits relatively cost-free offer of trial runs.... In a month of reading, Ican try out more "lives" than I cantest in a lifetime'.67 Poets from Wordsworth to Eliot have harboured similar ambitions,hoping to produce work that 'enlarges our consciousness or refines oursensibility'.68 Some such role has been played, from time to time, by social realist art,by photojournalism and by radio plays. Nowadays it is played most commonlyby what one critic dubs 'the imagination of the new media of directknowledge and graphic reproduction': film, video and television.69 My proposal is simply that we make use of those familiar phenomena forpurposes of constructing a model of deliberative democracy that isgenuinely feasible in large-scale societies. Instead of the (inevitably futile) attempt to make everyone else'communicatively present' in the same place at the same time, let usinstead try harder to make everyone else 'imaginatively present' in theminds of each of the deliberators. It might just be possible, through the exercise of a suitably informed imagination, foreach of us to conduct a suitably wide-ranging debate among all thesecontending perspectives largely within our own heads.

67Wayne C. Booth, TheCompany We Keep: An Ethics of Fiction (Berkeley: University ofCalifornia 68T. S. Eliot, 'The social function of poetry', On Poetry and Poets (London:Faber & Faber, 1958;

Press, 1988), p. 485.

originally published 1943), pp. 15-25. Elaborating onthat thought, Eliot writes that poetry conveys 'some fresh understanding ofthe familiar, or the expression of something we have experienced but have no words for, which enlarges our consciousness orrefines our sensibility' (p. 18); 'The genuine poet ... discovers newvariations of sensibility which can be appropriated by others. .... Inexpressing what other people feel he is also changing the feeling by making it more conscious; he is making people moreaware of what they feel already, and therefore teaching them somethingabout themselves' (p. 20). See similarly William Wordsworth, 'Observationsprefixed to "Lyrical Ballads"' (1820), What Is Art? ed. Alexander Sesokske (New York: OxfordUniversity Press, 1965), pp. 261-74. 69Steiner, 'Literature andpost-history', p. 420.

26 My suggestion, then, is that private fictions can serve public functions. Perhaps they cannot do so nearly as well as genuineone-on-one conversational exchanges could have done (though it pays toremember how much of the work of an ordinary one-on-one conversation isdone within each of the interlocutors' heads, anyway). But where society is not small enough to make genuineconversational exchanges among all the relevant public feasible,substituting internal debates for external ones might not be a badsecondbest. Certainly it seems at least as promising as all the other second-best solutions offered by other deliberativedemocrats for overcoming the constraints of large-scale society, as I shallnow go on to show.

V. Drawbacks of Internal Deliberation

Counting on 'internal-reflective' deliberation alone has some obviousdrawbacks. One is the obvious absence of an insistent 'other' who ispressing her perspective upon you.70 Some people and their perspectives might be ignored altogether; othersmight end up being more-or-less parodied because the too-pat representations of them we have inside our heads pass unchallenged.People whose situations are prototypical and familiar may be representedtolerably well in our internal deliberations; those whose situations arepeculiar in some way often will not.71 And so on.

70As Alan Ryan writes, 'In the absence of a real, physically presentinterlocutor who can interrupt,

question, rebut, shrug his or hershoulders, change the subject and at a pinch walk away entirely, you thereader are at the mercy of my ideas about what this conversation is about. ...[Y]ou cannot redirect the conversation as youwould wish'; 'In a conversational idiom', Social Research, 65 (Fall1998), 473-89 at p. 473. 71Michael F. Schober, 'Conversational evidence forrethinking meaning', Social Research, 65 (Fall 1998), 511-34. Seyla Benhabib, 'The generalized and theconcrete other: the Kohlberg-Gilligan controversy and moral theory', Situating the Self (Oxford: Polity, 1992), pp. 148-77. Note that it is not just the'peculiar' that presents a challenge, though: any departure from our ownway

27 All of that is true enough. That is why external-collectivedeliberation is superior to purely internal-reflective deliberation, wheresociety is sufficiently small to make external-collective deliberationgenuinely possible. But all thatseems much less of a worry where societies are of a size such thatgenuinely collective deliberation is not possible anyway. When comparingthe second-best solutions pressed upon us by the necessity of coping withlarge numbers, the internal-reflective mode of deliberation might be at least as good as any external mode that isthere actually available to us.

A. Attending to the Other

In substituting internal-reflective deliberation for external-collective,the first set of worries concerns who willbe 'included' in the deliberation. In externalcollective deliberations,deliberative democrats are at pains to ensure that everyone affected isparty to the deliberations; and they try to design the deliberative processwith a view to ensuring that.72 With internal-reflective deliberations, however, each deliberatorinevitably populates her own imaginary internal universe as she will. Of course we might exhort her to be as inclusive as she can, to tryvery hard to engage imaginatively with as many different sorts of people as might genuinely be affected by the decision. Wemight even send her a pile of books or photos or videos, as an aid to thatprocess. But there seems little that we can do from the outside to make the full range of others present to her mind's eye, in the

of thinking requires a stretch of the imagination which is, to some greater or lesser extent, difficult to achieve. 72See, e.g., John S. Dryzek, 'Political inclusion and thedynamics of democratization', American Political Science Review, 90(1996), 475-87.

28 way we mighthope to make all appropriate others physically present inexternalcollective deliberations of the more ordinary sort. Remember, however, the context of my present argument. I havesuggested we turn to internal-reflective deliberation precisely because (or, rather, insofar as) the group ofpeople affected is too large literally to make them all physically presentand still have a meaningful deliberation. Thus, what we should becomparing is, on the one hand, the representativeness of the population which we would conjure up in our mind'seye with, on the other hand, the (effective) representativeness of thesecond-best methods of externalcollective deliberation surveyed above. No doubt our imagination will always be imperfect, and some peoplewill be left out of deliberations based on internal-reflective processesalone.73 By the same token, however, those other second-best methods ofexternal-collective deliberation are imperfect too; and some people orpositions will always unrepresented or inadequately represented in them aswell. I see no way of settling a priori which mechanism will make the worst omissions. But it isnonetheless worth noting that they both run analogous risks in thisrespect. Note too that, even where others are physically present in externalcollectivedeliberations, that does not necessarily mean that we will be genuinelyresponsive to them and their concerns. We can always turn shrug ourshoulders, turn our backs or walk away.74 We canalways turn a blind eye or deaf ear.75 Input is no assurance of uptake. Indeed, recalling how heavily
73On the other hand, external-collective deliberations necessarily omit fromconsideration people

who cannot be present (future generations, in theexample which I elaborate in Section VI), whereas even non-existent(possible or probable future) people can be imaginatively present in internal-reflective deliberations. 74Ryan, 'In a conversational idiom', p. 473. 75As Averill Harriman famously did, ostentatiouslyswitching off his hearing aid when his Soviet counterparts launched intoone of their standard harangues. (I owe this anecdote to my old friend andteacher, Robert Ferrell.)

29 ordinaryconversation depends on internal representations, one might even say that'internal presence' is just as much a precondition of effective representation in external-collective deliberations as it is ininternal-reflective ones.

B. Understanding the Other

In real conversations between real people, there is a constantcross-checking and renegotiation of meanings.76 That facilitates the understandings which real interlocutors have of oneanother. People who are merely overhearing a conversation sometimes findit hard to understand what is going on, precisely because they cannotinterject into the conversation to cross-check their own understandings of what others mean to besaying.77 Whereas in real conversations a code of dyadically-shared meanings emerges,that simply cannot happen in imaginary conversations with imagined peopleof the sort that occur in the 'internal-reflective' deliberative mode.There, we are essentially havinga conversation with ourselves. If we are sufficiently imaginative, wemight envisage our 'imaginary other' correcting us in ways akin to those inwhich actual others might. But inevitably that is a pale shadow ofthe vigorous sort of cross-checking and cross-fertilization which occurs inany actual conversation. No single individual's imagination, however rich,will be able to

76'Conversation' can be defined 'as continuously negotiated communication';see Charles Tilly, 77As Sartre says, if we were to hear agramophone of an everydayconversation of a household in

'Contentious conversation', Social Research, 65(Fall 1998), 491-510 at p. 495.

Provins, we would not be able to understandit, lacking the background and context; JeanPaul Sartre, What IsLiterature?, trans. Bernard Frechtman (London: Methuen, 1950; originally published1948), p. 50. Formal experimental evidence in support of this speculationis provided by Michael F. Schober and Herbert H. Clark, 'Understanding byaddressees and overhearers', Cognitive Psychology, 21 (1989), 211-32; for a brief gloss, see Schober,'Conversational evidence for rethinking meaning', pp. 520-2.

30 mimic what occurs in the perfectly ordinary course ofevents in conversations among real people with genuinely different perspectives. Before we become too worried by that, though, let us reflect upon thefact that much the same is true of mass deliberations as well. In dyadicconversations, each speaker can sensitize the other to her own particular perspective one-onone. In mass deliberations, however, whattypically happens is that some speak and many listen. Hopefully those whospeak are broadly typical of many others who do not.78 But in any moderately large group there can be no realistic hope of eachperson individually negotiating meanings with each particular other,anyway. Thus, while 'internal-reflective' deliberations may look seriouslydeficient in the sorts of shared understandings they nurture when comparedwith conversational dynamics in dyads or small groups, once again that is not the relevantcomparison. The proposal here is to let internal-reflective deliberationssubstitute for external-collective ones in large groups. In settlings likethat, discursive dynamics are very different from conversational dyads anyway. There, the sorts ofintensely negotiated meanings which we find emerging in conversationaldyads will be largely missing. There, external representations risk beingjust as stylized, just as oriented toward the prototypical, as are the representations which figure in people'sinternal imaginary reconstructions of social life.79

78If only in theiratypicality: in a representative sample, the fact of diversity ought berepresented 79They inevitably reflect the 'generalized' more than the 'concrete' other,in all the other's concrete

even if not all the diverse components can be individuallyrepresented.

forms, in the terms of Benhabib, 'Thegeneralized and the concrete other'.

31 C. Representing the Other

As if in direct reply to my proposal to replace external deliberation withinternal, talking with reading, Montaigne protests: Studying books has a languid feeble motion, whereasconversation provides teaching and exercise all at once. If I am sparringwith a strong and solid opponent he will attack me on the flanks, stick hislance in me right and left; his ideas send mine soaring. Rivalry, competitiveness and glory willdrive me and raise me above my own level.80

Certainly we can sympathize with that sentiment. Playing chess againstyourself is far less satisfactory than playing against someone else (oreven a good computer): when playing against yourself, you always know whatthe 'other' is thinking; and that makes everything too pat, too devoid of surprise and of the creativity thatcomes from it. All that seems as true of cooperative games like conversation as it is of competitive games like chess. Furthermore, toparaphrase Mill, no one can know someone else's interests, position andperspective nearly so well as that person herself. For both those reasons,there is therefore a compelling case, pragmatic as well as symbolic, for the 'politics of presence': for alldifferent sorts of people to be physically present during deliberationsthat affect their interests, rather than just having their interests'represented' by others.81 Once again, however, that ideal seems to be compromised by therealities of large-scale societies. The circumstances here in view the circumstances

80Michel deMontaigne, 'On the art of conversation', The Essays of Michel de Montaigne, trans & ed.

M. A. Screech(Harmondsworth, Mddx.: Allen Lane/Penguin, 1991; originally published1580), Bk 3, essay 8, pp. 1044-69 at p. 1045. 81Anne Phillips, The Politics of Presence: Democracy & GroupRepresentation (Oxford: Clarendon Press, 1995).

32 under which I am recommending external-collectivedeliberation be replaced by internal-reflective are circumstances in which there are too many different people involved for all of them tobe effectively present in deliberations. At best, we might (through what I have called 'ersatz deliberation')substitute one or a few members of the group for all other members of thatgroup.82 Such a substitution might be satisfactory if the groups were sohomogeneous that a few members really were representative of the group as awhole. (In the limiting case if each were a literally identical token ofthe type, utterly interchangeable with every other, then a single representative from each group would suffice.)Most theorists of group difference would baulk at the sort of'essentialism' implied by that, however.83 They shun the'generalized' in favour of the 'concrete other'.84 And oneimplication of that is that no small set of representatives can truly standin for groups as a whole, as the politics of presence would require. How much this will impinge on external-collective deliberative processesdepends in part on how many different groups there are to be representedand in part on how many individuals it takes to represent tolerably welleach of those groups in its full complexity. But in any large-scale society, we can probably assume moderatelyhigh levels both of inter-group pluralism and of intra-group heterogeneity. And that multiplicitywould be compounded yet again, if geographical situatedness itself provesto be an important dimension of identity and difference, as theorists ofthe 'politics of place' contend.85

82What Phillips's calls for is precisely that: 'group representation', asher subtitle makes clear; see 83Young, 'Together in difference'. 84Benhabib, 'The generalized and the concrete other'. 85J. E. Malpas, Place & Experience (Cambridge:Cambridge University Press, 1999).

esp. Politics of Presence, ch. 6.

33 The upshot would seem to be, once again, that in any large-scale societyexternal-collective deliberations are necessarily very far from ideal.That deliberation cannot effectively give every distinct voice a hearing:second-best short-cuts of one sortofanother will inevitably be required; and something will inevitably be lostin the process. It is once again an open question how much would be lostthere, compared to how much would be lost by trying instead to replicatesome such conversation wholly within one's own mind in an internalreflective deliberation. But once again,at least we can say that the clear advantage that external-collectivedeliberation enjoys over internal-reflective deliberation in the ideal caseis clearly eroded, in the case actually at hand.86

D. Finding Time for the Other

Finally it might be argued that, in any large-scale mass society,internal-reflective deliberative processes would fall prey to the samepressures on time and attention as do external-collective deliberativeprocesses. In the latter case, the problem is that we lack the time to have the requisite conversations with allothers. In the former case, it might be said, there is a perfectlyparallel problem: we lack the time to imagine those conversations,either. Certainly it is true that 'attention' is a strictly limited resource,imposing severe constraints on our deliberative capacities. Herbert Simonreports: The human eye and ear are highly parallel devices, capable of extractingmany pieces of information simultaneously from the environment and decodingthem into their significant features. Before this

86True, in situations of great heterogeneity we might find it hard to imagineourselves in a very

different other person's position: but thatcompromises our capacity for understanding what the other is claiming, inexternal-collective deliberations, just as much as it compromises our capacity for imagining ourselves her forinternal-reflective deliberative purposes.

34 information can be usedby the deliberative mind, however, it must proceed through the bottleneck of attention a serial,not parallel, process whoseinformation capacity is exceedingly small. Psychologists usually call thisbottleneck short-term memory, and measurements show reliably that it canhold only about six chunks (that is to say, six familiar items ofinformation). .... The narrowness of the span of attention accounts for a great deal of human unreason thatconsiders only one facet of a multifaceted matter before a decision isreached.87

Those limits on human cognitive capacities, however, impose no extra constraints on internal-reflective deliberative processes as compared to externalcollective ones. Inexternal-collective deliberations, just as surely as in internalreflectiveones, we must find time to attend to all the persons and perspectives thatare present. Indeed, we have to find more time for them, there: the mechanism by which we attend to others throughexternal-reflective processes is oral or written communications, whichmeans that we can attend to only one of the others at a time; and thatmakes the external-collective deliberative process a more radically serial process than isinternal-reflective deliberation. Even if we can only effectivelymulti-track six different things at once in our internal reflections, wecan nonetheless 'listen' to five more people/perspectivesat once through that internal-reflective process than we can through theexternalcollective one. Finally, suppose we manage successfully to 'internalize' theperspectives of various others, through having imaginatively projectedourselves into their position on some previous occasion. Then perhaps we might even be able to'see' the situation from those many different perspectives at once withoutany conscious effort. If those other perspectives have been internalizedin some strong way, applying them is 'second nature' to us. No deliberate act of will is

87Herbert A. Simon, 'Human nature in politics: the dialogue of psychologyand political science',

American Political Science Review, 79 (1985),293-304 at p 302.

35 required to evokethem; no deliberate focusing of our attention on them is involved.'Seeing' things from all those other perspectives might then be more likethe 'parallel' processes which Simon describes governing the eyes and the ears than it is like the 'serial' processof consciously directing attention first here and next there. Again, there is no guarantee that people will 'internalize' any (muchless all) other relevant perspectives in this strong way.88 The point is merely that they might. If they do, that wouldsignificantly ease the cognitive constraints involved in attending to manyothers, in the case of internal-reflective deliberation as compared toexternal-collective. But even if they do not, internalreflective deliberation is no worse off (and arguablybetter off, by a factor of sixto-one) than external-collectivedeliberation, with respect to those cognitive constraints.

VI. The Democratic Imagination

My argument, in short, is that internal, imaginary discourses can beimportant sites of democratic deliberation, particularly in large-scalesocieties where the scope for genuinely deliberative external-collectivediscourses are so limited. Recognizing that fact has important practical consequences. It makes central concerns which might otherwise seemperipheral to our democratic theory, concerns to do with the production andconsumption of the representations and images upon which our imaginingswork. On the one side are familiar questions about access to means of modern mass communication.

88And if they internalize some but not all other perspectives, there is ofcourse then a risk that

their internal-reflective deliberations will bebiased accordingly.

36 On theother side are equally familiar questions about 'for whom does onewrite?'89 Who is the audience, the reference group?Who are the subjects, and how are they represented?90 Thisis not the place to launch into those familiar debates in literary theory.I merely hope to have shown that, far from being arcane disputes withinliterary and cultural studies, those are actually issues which areabsolutely central to deliberative democracy as it is inevitably practiced in modern largescale societies. There,representations inside the head count for at least as much asrepresentation in any legislative chamber. And the art forms out of whichwe construct those representations are potentially as potent, politically, as are the elections out of which weconstruct legislative assemblies. It is not a novel thought that democracy requires well-stocked publiclibraries and public funding of the arts.91 It is not a novel thought that 'artistic creations shape politicalconceptions', or that public museums should be 'forums, nottemples'.92 All of those familiar debates obviously bear on the topic I am heretrying to get onto the table of the democratic theorist. But for the mostpart they bear only obliquely. Thus, it is true and important that art is in part a public good, andit would be undersupplied in consequence by ordinary market forces; weought subsidize

89Sartre, What Is Literature? ch. 3. 90In terms of democratizing culture (or, rather, of enlisting culturalartefacts in the service of

democracy), it is not so much a matter of 'highculture' against 'low' as it is of 'broad' against 'narrow'. 91Amy Gutmann,Democratic Education (Princeton, N.J.: Princeton University Press, 1987), ch. 8. Dick Netzer,The Subsidized Muse: Public Support for the Arts in the United States(Cambridge: Cambridge University Press, 1978). Edward C. Banfield, The Democratic Muse: The Visual Arts and the Public Interest (New York:Basic Books, 1984). Carnegie Commission on the Future of PublicBroadcasting, A Public Trust (New York: Bantam, 1979). 92Murray Edelman, >From Art to Politics: How Artistic Creations Shape Political Conceptions(Chicago: University of Chicago Press, 1995). Robert McC.Adams, 'Forums, not temples', American Behavioral Scientist, 42 (#6:Mar 1999), 968-76.

37 creativity, insofar as we can.93 And, to shift from the production to the consumption side, it is true andimportant that we ought 'take art to the people', rather than locking itaway in the Ivory Tower. All of that is crucial in generating anddisseminating the sorts of representations which will serve as crucial aids to the sorts of internal-reflectivedemocratic deliberations that I am here proposing. Equally importantly, though, we need to ensure the representativeness ofthose representations. What in effect we will be doing, in shiftingdemocratic deliberation from the external-collective to theinternal-reflective modes, is enfranchising images.In so doing, we obviously need to ensure that the images thus enfranchisedare as extensive as required really to represent the diversity ofexperiences extant across the communities to be affected by thosedeliberations.94 Not all of those experiences are pretty. Not all are intellectuallyedifying or morally uplifting. Some will be sad or depressing or downrightobnoxious. Still, all deserve a voice in the democratic cultural space,insofar as that feeds into internal-reflective deliberation or anyway all do, to just the same extentthat they deserve a literal voice in the democratic political space that issupposed to feed into external-collective deliberations. Insofar as we have good democratic grounds for censoring 'hate speech', we might beprepared to ban its other cultural manifestations as well.95 But from the present perspective, we have no

93Ronald Dworkin, 'Can a liberal state supportart?', A Matter of Principle (Cambridge, Mass.: 94Issues of 'neutrality' emerge in state funding of the arts, insofar as notall projects can be funded.

Harvard UniversityPress, 1985), pp. 221-33.

Strictly analogous issues arise inconsultative fora, insofar as not all interested parties can be given aseat at the table. The solution, in both cases, is to ensure that the subset which is funded/seated is genuinelyrepresentative of the larger set from which it is drawn to ensure that all distinctive perspectives are adequately represented.Cf. Harry Brighouse, 'Neutrality, publicity and public funding of thearts', Philosophy & Public Affairs, 24 (1995), 3663 and Dworkin,'Can a liberal state support art?' 95For examples see Karl Lowenstein, 'Legislative Control of PoliticalExtremism in European Democracies', Columbia Law Review, 38 (1938):591-622 and 725-74.

38 more grounds for confiningour public concern, and public subsidies, purely to more 'elevated' formsof cultural expression than we do for confining our political attentionpurely to the expression of 'elevated' opinion. One final advantage of internal-reflective deliberations is that,precisely because they do not require people to speak for themselves, they might hope to secure better representationof the communicatively inept or the communicatively inert thanexternal-collective deliberations ever could. Consider as a limiting casefuture generations.96 Our actions and choices today clearly affect them, and according toordinary democratic canons everyone affected ought have a say in ourdeliberations. But the unborn, by nature, cannot speak for themselves; andothers who purport to speak for them will inevitably be asked bywhat authority they do so. Internal-reflective deliberations experience nosuch problems. They do not need future people to be physically present inorder for them to be imaginatively present; and the whole process proceedsby everyone imagining themselves into the place of others, no special warrant is required for each of usto imagine ourselves into the place of people in the future.97 At some point, collective action and hence some collectivedecision is going to be required. For that, we must merge all the products of all our private internal-reflective deliberationsinto some common collective determination, somehow. In that sense,internal-reflective deliberations are thus not a subtitute for but ratheran input into external-collective ones. My point is merely that the

96Peter Laslett and James S. Fishkin, eds, Justice BetweenGenerations. Philosophy, Politics & Society,

6th series (Oxford: Blackwell, 1992). Other examples of communicativelyinert interests which we arguably ought take into account might includeinterests of other species, people in distant countries and so on; RobertE. Goodin, 'Enfranchising the earth, and its alternatives', Political Studies, 44 (Dec. 1996), 835-49. 97I am grateful for helpful comments and criticisms from Louise Antony,Martin Davies, John Dryzek, Jim Fishkin, Dick Flathman, Martha Nussbaum andPhilip Pettit.

39 more democratically deliberative our internalreflections are, the less worry there needs be about the impossibility ofour external collective decision procedures being particularlydemocratically deliberative in large-scale mass societies.