Beyond the Fact of Disagreement? The Epistemic Turn in Deliberative Democracy

Abstract

This paper takes stock of a recent but growing movement within the field of deliberative democracy, which normatively argues for the epistemic dimension of democratic authority and positively defends the truth-tracking properties of democratic procedures. Authors within that movement call themselves epistemic democrats, hence the recognition by many of an ‘epistemic turn’ in democratic theory. The paper argues that this turn is a desirable direction in which the field ought to evolve, taking it beyond the ‘fact of disagreement’ that had previously blocked the conceptual road to acknowledging more than intrinsic properties to democratic decision procedures. The paper shows how two authors in particular – Joshua Cohen and David Estlund – have successfully lifted the Rawlsian requirement of epistemic abstinence and defends epistemic democrats and the implications of the epistemic turn in democratic theory against various misconceptions.

Keywords:

• Epistemic democracy
• deliberation
• disagreement
• truth
• public reason

Notes

1. The distinction between knowledge-aggregation and truth-tracking is that the ability of a procedure to aggregate knowledge does not strictly guarantee its ability to track the truth, though the two abilities are likely highly correlated. There is, also, a distinction between the property of tracking the truth (however defined) and the ability to produce good outcomes, empirically speaking (except perhaps on a pragmatist conception of the truth where the truth is defined as ‘what works’ and is thus deeply connected to a certain type of consequences in the world).

2. In this issue Chambers rightly emphasizes the often unappreciated epistemic (as well as systemic) dimension of Habermas’s early work. I do not entirely follow her, however, when she more generally suggests that ‘the core principle of deliberative democracy [is] that epistemic quality and equal participation are connected’ (Chambers 2017, 5). To my mind the core principle of deliberative democracy is that the legitimacy of the law derives from having been the object of public deliberation among free equals. The distinct thesis about the relationship between equality and epistemic quality was not really made explicit, let alone argued for, until the work of epistemic democrats, starting with Estlund’s (1997) seminal article. That this distinct thesis can now be claimed and endorsed by deliberative democrats as compatible with their core statement is of course an excellent thing.

3. Taking the epistemic framework for granted, some authors are now exploring the epistemic constraints on citizens’ judgments, such as framing effects and other cognitive biases (Kelly 2011).

4. Building together on their separate work (e.g. Misak 2004; Talisse 2009).

5. That is how I read them anyway; see also MacGilvray (2014). In my view, Zerrilli’s recent book (2016) also seems to tread into epistemic territory by emphasizing the objective, ‘world-building’ (if not quite truth-tracking) properties of democratic judgment over the radical subjectivism of earlier Arendtian interpretations.

6. This problem is less accurate in deliberative theories of a Habermasian vein, because they start not from the fact of pluralism as a feature of modern societies, but from the common structure and pragmatic conditions of possibility of human discursive exchanges. Even in Habermas and his followers, however, the epistemic dimension of deliberation, always latent, has only recently been made explicit (Habermas 2006; Buchstein and Jörke 2012; and Jörke in Fischer and Gottweis 2012, 277).

7. An overlapping consensus is one in which people simply come to support identical principles or conclusions on the basis of compatible but distinct, and more or less comprehensive views or premises. This is in contrast with a ‘rational consensus’ where individuals agree on all premises and conclusions. Note that the domain of applicability of Rawls’ argument is rather limited – constitutional matters regarding the basic structure of society – but a lot of deliberative democrats apply the ideal of the overlapping consensus to debates about less fundamental legislation as well.

8. They include the conflicting, ambiguous, and complex nature of the evidence bearing on any interesting case; the differences in the weight people attribute to factors or considerations even when they agree on them; the vagueness of all our moral and political concepts; the diversity of our experiences, which shape our diverse perception and understanding of the world; the difficulty of making overall assessments about multi-dimensional problems; and the ‘limited social space’ for accommodating a diversity of values, which makes it necessary and yet impossible to agree on the right ranking.

9. The reason why Rawls thinks that human cognitive fallibility should be integrated into ideal theory, while moral fallibility should not, for example, is that one should expect of ‘reasonable’ human beings some self-control over their immoral or selfish impulses, while one cannot expect even reasonable beings to change the world or fix their psychological and neurological wiring.

10. In the passage cited above, reasonable pluralism was tellingly introduced as an ‘idea.’ Later Cohen presents this idea as a mere ‘contention’ that is ‘suggested by the absence of convergence in reflection on [all sorts of] issues of value’ (Cohen 1993, 282, my emphasis).

11. This choice is, naturally, fully conscious. It would run against Waldron’s democratic commitments to endorse an a priori criterion of the ‘reasonable’ selecting at the gate of deliberation who and which views count as such.

12. For example, Gutmann and Thompson’s characterization of some disagreements as ‘deliberative’ and other as ‘not-deliberative’ (and therefore not really acceptable) maps obviously and directly onto Rawls’s reasonable pluralism as reconstructed by Cohen (e.g. Gutmann and Thompson 1996, 3). Yet, some of the examples of significant moral disagreements they give do not bracket, as Rawls’s ideal theory does, the possibility of self-interest, as in the conflict between ‘the president and the reporter’ about ‘what fairness requires,’ or that between ‘the AT&T executive and the employee’ about ‘what justice in employment means, as well as about who should suffer its costs’ (1996, 15). In fact, Gutmann and Thompson seem to accept disagreement that stems from other non-reasonable sources. For example, they count ‘simple misunderstanding’ as an illustration of the persistence of disagreement. Though it is not entirely clear in what a ‘simple misunderstanding’ consists, it would appear to be the kind that could be avoided with the extra attentiveness or focus that reasonable people in Rawls’ story are assumed to provide.

13. Arguably, more empirically-minded deliberative democrats, like Jim Fishkin, always explicitly cared about the way a certain type of deliberation may remove or avoid the primary suspects for epistemically poor decisions, such as polarization, bad heuristics, and biases of all kinds (e.g. Fishkin 2009). This should have made them good candidates for seeing epistemic democracy as a framework with which to make sense of their questions about the impact of deliberation. At most though, Fishkin is willing to suggest that post-deliberative opinions are likely more ‘informed’ than pre-deliberative ones but he remains agnostic as to whether deliberation is likely to bring us closer to any kind of procedure-independent standard.

14. In an earlier text, Rawls had already resisted a similar claim that reaching a reflective agreement on a conception of justice should be ‘sufficient grounds for regarding that conception as true, or at any rate highly probable.’ As he forcefully put it then, ‘we refrain from this further step: it is unnecessary and may interfere with the practical aim of finding an agreed public basis of justification’ (Rawls 1987, 15, my emphasis).

15. I rely here on the views expressed by Cohen and Estlund in two essays conjointly published (or in Cohen’s case republished) in the recent edited volume by Norris and Elkins Truth and Democracy (2012) The references to Cohen are to the original (2009) publication. I leave Raz’s (1990) famous critique outside of this paper because on my reading Raz ends up offering an external critique of Rawls, whereas Cohen and Estlund try as much as possible to rehabilitate truth from within the Rawlsian framework.

16. See Cohen (2009, 26) for the details of this conception of justice, which is closest to, though importantly distinct from, the deflationist or so-called disquotational theory of truth, whereby the proposition that p is true if and only if p is true (Cohen 2009, 4).

17. Estlund’s solution may well be more epistemic than Estlund himself is willing to admit since, on his account, truth unavoidably sneaks through the back door not just at the level of the individual exchange of arguments but at the level of the legitimizing procedure as well. Democratic deliberation is not supposed to always produce the ‘true’ answer but it is legitimizing only as long as it is likely to produce truer answers, or true answers more often, than a random procedure. So it is not the case, even on Estlund’s terms, that all we need at the level of the legitimizing procedure is the Rawlsian concept of the ‘reasonable.’ What we need, it seems to me, is the concept of truth.

18. Landemore makes similar points (Landemore 2013, xvii–xviii, 213) and proposes a typology of political cognitivisms, from the more absolute and universalist to the more culturally relative and minimalist (Landemore 2013, Chapter 8).

19. As they put it: ‘That we could not recognize our normative beliefs except as truth-aspiring and reason-responsive grounds a cognitive conception of moral and political discourse’ (Misak and Talisse 2014, 3, their emphasis).

20. Or as he also writes, ‘One need not be any kind of Platonist in order to entertain epistemic justifications. All that is required is that there is some desired procedure-independent property, so that one can intelligibly ask whether a procedure tends to yield decisions with the desired property’ (Ingham 2013, 139). What he disagrees with is the view that one may ever agree on the nature of the property at stake.

21. Recent research has tried to codify the properties of good deliberation. See for example the Discourse Quality Index (DQI) developed by Steenbergen et al. (2003); the index of intersubjective consistency by Niemeyer and Dryzek (2007); the index of ‘cognitive complexity’ by Wyss, Beste, and Bächtiger (2015); the measure of adequate support for given conclusions by Schaffer and Friberg-Fernros (2017) or the consideration for validity and scope in Knops (2017).

22. Impact assessment is a new policy trend that seems rather commonsensical, at least in some areas where outcomes lend themselves to empirical measurement.

23. For the classic distinction between justification and legitimacy, see Simmons (2000). Among authors who take this distinction seriously, only David Estlund, however, seems to believe that epistemic considerations, usually considered meaningful for the context of justifying democracy, should also enter the question of its legitimacy (for skepticism on that front though, and why epistemic considerations may have an easier time justifying an institution’s existence than justifying legitimate authority, see Viehoff [2016]).

24. Truth remains present indirectly though, via the procedural pedigree of the laws that are being enforced (it is after all because people thought certain policies or laws were the correct ones that they voted for them and that they now can be legitimately enforced). Thank you to Daniel Viehoff for this point.

25. See also on the risk of epistocracy: Føllesdal (2017), Santoro and Liveriero (2017), and Holst and Molander (2017).

26. The central argument builds on the epistemic properties of inclusive deliberation as arguably captured by formal results such as Scott E. Page’s Diversity Trumps Ability Theorem (Page 2007). Inclusive deliberation benefits from the cognitive diversity of the entire group, as well as the lesser but nonetheless important epistemic properties of judgment aggregation (which are themselves captured by results like the Condorcet Jury Theorem, the Miracle of Aggregation, or the Diversity Theorem). See Landemore (2013) for more details.

27. See also Lafont (2014) for a similar point about the value of randomly selected ‘mini-publics’.

28. That is, I believe, what Knight and Johnson more radically suggest when they argue that political equality can only be justified on the grounds of the pragmatic benefits it yields (Knight and Johnson 2011).