WP6. Epistemic Risk in Collective Contexts

There are at least two types of epistemic risks that specifically arise in collective contexts. We have called the first one individual-group dissonance risk, and we have conceptualized it as twofold risk that affects both individual group members and groups as a whole. In particular, by being member of a group an individual may be at risk that her beliefs outside the group clash with the group's view, which is best understood in terms of Gilbert's plural subject account: the relevant tension is between the individual group member’s beliefs and her participation in a joint commitment to accept opposite or inconsistent views as the group's view. On the other side of the coin, collective agents run this sort of epistemic risk when individual group members could easily form a joint commitment to let a certain view stand as the group’s view that none or few of them individually accept outside the group—non-summativism and in particular Gilbert's plausible plural subject account allow for this possibility. WP6 will investigate the epistemic significance as well as the epistemological consequences of individual-group dissonance risk in the context of the (very recent) discussion on epistemically justified group belief (see Goldman 2014; Lackey 2016).

The hypothesis to explore is that, when it comes to the question of what a justified group belief is, there are two families of competing approaches that mirror the two main approaches to individual epistemic justification in the literature: internalism and externalism. At least two internalist views will be taken into account: the view that group justification is a matter of groups responsibly responding to their total evidence, i.e., evidentialism about justified group belief (Silva forthcoming; also Lackey 2016), and the view that a group belief is justified if and only if the beliefs of individual group members are consistent (for a suitably understood notion of consistency) both with each other and with the relevant collective belief, i.e., coherentism about justified group belief. One strand of argument that will be explored against these views is that, first, the risk of individual-group dissonance, when it materializes, is incompatible with the relevant group beliefs being epistemically justified and, second, that a group can comply with the relevant internalist standards for epistemic justification without excluding this sort of epistemic risk. In addition, it will be argued that externalist theories (e.g., Goldman 2014), by modeling group justification in terms of conditional reliability (i.e., a group’s belief is justified only if the beliefs of its individual members are justified, where the justification of the collective belief comes from a reliable procedure for aggregating individual beliefs) are better positioned to deal with this sort of epistemic risk than internalist views.

The second form of epistemic risk that arises in group settings is the risk that deliberation becomes unreliable for forming group attitudes due to certain group dynamics: group polarization, hidden profiles, and the amplification of individual cognitive errors (deliberative epistemic risk). WP6 will investigate the epistemic significance as well as the epistemological consequences of this sort of risk in the context of (the also very recent) discussion on the epistemology of group disagreement, which investigates the rational response to disagreement (e.g., about politics, religion, social issues, factual matters) in group settings—see Broncano-Berrocal & Carter (forthcoming) for an edited volume that compiles most work carried out on the topic; Carter (2014) and Skipper & Steglich-Petersen (forthcoming) for two recent contributions; Frances & Matheson (2018) for an overview of the more standard literature on the epistemology of peer disagreement. Related to this, it is a common phenomenon that a group holds a view that results from a consensus that its members reach following an internal disagreement. Crucially, disagreeing and posterior agreeing on whether a certain view should stand as the group’s position can be a deliberative or a non-deliberative process. While deliberation is generally considered more epistemically desirable than non-deliberation (e.g., because it may help distribute the evidence evenly), empirical research shows that it can yield epistemic bad consequences: individual cognitive errors can be amplified, group consensuses can become more extreme (group polarization), or evidence can be concealed (hidden profiles).

WP6 will address two connected questions vis-à-vis the realistic possibility of deliberative epistemic risk: (1) what is the most epistemically respectable way to solve intra-group disagreement, deliberatively or non-deliberatively? More specifically, is deliberation more epistemically advantageous or disadvantageous than non-deliberation when it comes to solving intra-group disagreement by reaching group consensus if it is reached in conditions of deliberative epistemic risk? (2) Which conditions should deliberative disagreement comply with to be epistemically proper? More specifically, what it would take to eliminate deliberative epistemic risk and thus overcome or at least mitigate its epistemic disadvantages when it comes to deliberative intra-group disagreement?

To answer the first question, WP6 will review and discuss current empirical evidence to compare how well the deliberative and non-deliberative forms of intra-group disagreement fare with respect to familiar epistemic standards such as reliability in producing accurate group consensuses or conduciveness to certain epistemic goods (other than truth) that are related to collective epistemic virtues. The working hypothesis will be that there is no clear-cut answer because the two options exhibit different epistemic advantages and disadvantages. Thus, depending on one’s epistemic interests and preferred epistemic goals, one will deem one of the options more epistemically respectable than the other.

With this caveat in mind, WP6 will address the second question by resorting (once again) to Gilbert’s plural subject account of group phenomena and more specifically to her notion of joint commitment. The working hypothesis here will be what a group can do to overcome or mitigate the epistemic disadvantages of deliberative and non-deliberative disagreement is that its members make certain joint commitments (which will be specified) so that disagreement and its subsequent resolution can take place in conditions that respect the relevant epistemic standards.

Finally, WP6 will tackle some ethical issues concerning deliberation. In particular, the key question that WP6 will try to answer is this: given the epistemic limitations of deliberation, are there epistemic reasons for banning or, less dramatically, constraining deliberation in some contexts? For example, Hedden (2017) has recently argued that the various deleterious effects of jury deliberation jointly justify a mandatory non-deliberation policy for jurors that limits the information available to jurors and constrains their decision-making powers, with the aim of increasing accuracy. But in stark contrast with his, deliberation generates various kinds of clear epistemic benefits such as pooling of evidence (Hong & Page 2012) or positive effects on juror memory (Vollrath et al. 1989). WP6 will assess existing arguments for and against constraining deliberation and will implement them in the context of the epistemology of group disagreement, and especially in connection with deliberative epistemic risk.