The impact of cognitive bias on deliberation

While there is significant empirical evidence to suggest that participants can become more informed, change their opinion and become more “pro-social” as a result of deliberation, there is also a large body of research that suggests that individuals may also do the contrary – entrench and make the available information fit with their pre-existing and self-serving opinions.

Mendelberg reports that: ‘In a literature known as motivated reasoning, social and cognitive psychologists have documented the variety of innovative ways that people who are strongly committed to a predetermined view find to interpret evidence to support their view. This bias occurs at every step of information processing, from setting goals, to gathering and evaluating evidence from the outside or from memory, to constructing inferences and judgments […]. People not only fail to attend to evidence that disconfirms their view, but they readily accept evidence as valid if it agrees with their view while questioning and ultimately rejecting the validity of information that challenges it […].’

Confirmation bias

This phenomenon of people taking information to fit pre-existing beliefs is commonly referred to as confirmation bias. In a famous study from the behavioural economics literature, participants with strong views on capital punishment were given research on its impact as a deterrent. Rather than using the evidence to challenge their pre-existing beliefs, not only did participants use it to reinforce their prior views, they used the language and reasons from the research to support their argument.

Some research into the effects of Deliberative Polling supports this finding, suggesting that the more salient an issue is, the less likely participants in a deliberation are to change their views on it:

If respondents have already processed an issue, even with fairly imperfect and unbalanced deliberation in their daily lives, they are less likely to change their views. 1

As discussed earlier, reinforcing existing views is not necessarily an illegitimate outcome of deliberation. Indeed, it might be of even more concern if participants were found to change their pre-existing views unreservedly due to deliberation. However, this does challenge the traditional theoretical underpinnings of deliberative engagement that emphasise rationality and the salutary effects of deliberation on participants’ opinions:

Motivated reasoning has considerable power to interfere with the motivation that deliberative theory cherishes the motivation to be open-minded, even­handed and fair. Deliberators can hardly pursue truth and justice if they view everything in favour of their priors through rose-tinted glasses and everything against it through dark ones. 2

However, rather than undermining the practice of deliberation itself, this might instead suggest that we need a more nuanced view of the purposes and outcomes that deliberation can achieve under different conditions. Theorists have recently suggested the need to combine deliberation with other forms of democracy in order to deal with irresolvable conflicts in opinions and interests. We should not mislead ourselves into thinking that those designing a deliberative process can control ‘the whole process’ from start to finish, because people will engage with the issues before and after a dialogue in ways that are outside the facilitators’ control.

Rather than imagining dialogue processes as hermetically sealed units that citizens can come to fresh, we should actively acknowledge citizen engagement that pre-dates a dialogue, and that participants may pursue other pathways for democratic voice after dialogue processes have finished.

Can deliberation overcome cognitive bias?

Some studies suggest that the qualities of a deliberative process may overcome, at least in part, the effects of cognitive bias. Barabas, for example, argues that the obligation on citizens participating in a deliberation ‘to open up to the possibility of attitudinal change’ and the structure of deliberation that means ‘people with diverse views commingle’ mean that deliberation is different from typical discussion. 3  Through his research into how a deliberative forum affected policy opinions, he finds that, ‘deliberation increases knowledge and alters opinions, but it does so selectively based on the quality and diversity of the messages as well as the willingness of participants to keep an open mind’. Specifically, Barabas finds that greatest opinion changes result from groups that find a significant degree of consensus on an issue, and concludes that:

Keeping an open mind, along with exposing ourselves to new information and diverse perspectives, is the essence of deliberation. It is what separates deliberation from discussion and other opinion influences. In two different studies, I have shown that citizens learn when they deliberate but not when they discuss politics. Deliberation is unique in that citizens discard their inaccurate factual perceptions as well as rigidly held political views. Deliberation represents an opportunity for opinion change, in the spirit of enlightenment and consensus, but there are no guarantees.  3

An evaluation, conducted by Esterling and colleagues, 4  of a deliberative event on long term planning for the US federal budget, presents similar findings. The deliberative process, which was held over a day in June 2010 across 19 communities, involved 3,000 individuals from a cross section of society, in discussing how to balance the US federal budget. The concern/expectation, based on the motivated reasoning literature, might be that liberals and conservatives would leave the deliberation with more extreme opposing positions than when they arrived. In fact, what was found is that liberals, conservatives and neutrals were able to take part in a constructive discussion and moderated their policy views on spending cuts and tax increases. The evaluation suggests that this was, at least in part, due to the organisers being able to create a forum for open and balanced discussion.

The authors also propose that another force at work that encouraged the moderation of individuals’ views was the common objective of agreeing a strategy to reduce the deficit. They observe that:

On different policy items, liberals and conservatives seem to have given ground on their specific priorities in order to help achieve this goal over the course of deliberation. For example, conservatives became more supportive of raising taxes on the very wealthy (liberals began with high levels of support for this measure and didn’t change much). To a similar degree, liberals became more supportive of a 5% across the board cut to discretionary programs after one day of deliberation. 4

The success of this deliberative engagement on such a value-laden and emotive subject is surprising considering that a significant amount of evidence (e.g. on group polarisation) suggests that cognitive biases and social pressures are most prevalent on issues of value. This literature suggests that deliberative engagement is best suited for issues that are technical, and do not require the discussion of competing values between groups of participants. Mansbridge goes so far as to argue that:

On matters of value, opportunities for deliberation are likely to tum anti-deliberative. And if they manage to tum argument-centered, they are unlikely to change minds. Advocates of deliberation would do well to promote deliberation on issues of fact but to advance alternatives to deliberation on issues of value. 5

The question of how an issue is framed, however, might go some way to explain how cognitive biases can be overcome even on issues that are heavily values laden. Returning to the example, the evaluation observes that the small group discussions were not strongly ideological in structure. Participants are reported to have been nearly unanimous that discussions were constructive and engaging, even where they were in groups with people very different from themselves. This would suggest that it is possible to hold a constructive deliberation on what would typically be a valued-laden subject when discussions are facilitated in the right way (e.g. avoiding ideological conversations and working towards a common objective). Esterling and colleagues, the evaluation authors, conclude:

When asked to discuss policies with their fellow citizens, participants tended to set aside their ideological commitments to work toward the common goal of fiscal responsibility. If one were to rely exclusively on individual survey responses to gauge public opinion, one would be misled to believe that our society can only consider policy options through a rigid ideological lens. But public opinion surveys have their limits in helping us understand the structure of public opinion. Public deliberation helps to reveal the considered opinions of citizens, a kind of opinion policy makers should care about as well. 4

Further evidence reported by Mendelberg suggests that the diversity of a group has an important bearing on whether it falls victim to cognitive biases; while homogenous groups will typically use information to confirm their pre-existing views, heterogeneous groups are less prone to this:

Groups, especially if they are homogenous, are much more prone than individuals to search for information that confirms their preliminary preference […]. One group mechanism that exacerbates the individual tendency to search for confirming evidence is the group’s ability to heighten individuals’ “defense motivation” – the feeling that once one has made a decision, one should commit to it. Homogenous groups also work by increasing members’ confidence; when a group agrees on what to do, the members are much more confident in that decision than they would be if making the same decision or when the group fails to agree […]. Heterogeneous groups are much less susceptible to these group biases. 4

The design of deliberative process, therefore, can have a significant bearing upon the degree to which cognitive biases are present and how individuals approach a discussion. The perennial concern to ensure a balanced and appropriately mixed group of participants in dialogue is therefore not only important in terms of ‘representativeness’, but also for minimising the risk of cognitive bias. Without such careful design, the impact of being part of a dialogue process may be to entrench citizen views in biased positions.  Conversely, with a well-run and carefully balanced group, dialogue can be a means whereby biases are addressed and corrected.

  • 1Fishkin, 2009, When the people speak
  • 2Mendelberg, 2002, The deliberative citizen: Theory and evidence
  • 3 a b Barabas, 2004, How Deliberation Affects Policy Opinions
  • 4 a b c d Esterling et al., 2010, The Difference that Deliberation Makes
  • 5Mansbridge, 1983, Beyond adversary democracy.