Is cultural cognition an instance of “bounded rationality”? A ten-yr debate

This is basically what I remember saying last week at William & Mary in a workshop co-sponsored by the Law School & Political Science Dep’t a couple weeks ago. Slides here.

1. An old but continuing debate.  The paper you read for this workshop—Motivated Numeracy and Enlightened Self Government, Behavioural Policy (in press)—originates in a debate that started 10 yrs ago.

A group of us (me, Paul Slovic, Donald Braman, and John Gastil) had written a critique of Cass Sunstein’s then-latest book Laws of Fear.  In that book, Sunstein had attributed all manner of public conflict over risk to the public’s overreliance on “System 1” heuristic reasoning. The remedy, in Sunstein’s view, was to shift as much risk-regulatory power as possible to politically insulated expert agencies, whose members could be expected to use conscious, effortful “System 2” information processing.

Our response—Fear of Democracy: A Cultural Evaluation of Sunstein on Risk, Harvard L. Rev., 119: 1071-1109—criticized Sunstein for ignoring cultural cognition, which of course attributes a large class of such conflicts to the impact that cultural allegiances play in shaping diverse individuals’ risk perceptions.

The costs of ignoring cultural cognition, we argued, were two-fold.

Descriptively, without some mechanism that accounts for individual differences in information processing, Sunstein could not explain why so many risk controversies (from climate change to gun control to nuclear power to the HPV vaccine) involve conflicts not between the public and experts but between different segments of the public.

Prescriptively, the cost of ignoring cultural cognition undermined Sunstein’s central recommendation to hand over all risk-regulated decisionmaking to independent expert risk regulators. That recommendation presupposed that all disagreements between the public and experts originated in the public’s bounded rationality, a defect that it was reasonable to assume could not be remedied by any feasible intervention and that generated factual errors unentitled to normative respect in lawmaking.

Cultural cognition, we argued, showed that public risk perceptions on many issues were rooted in diverse citizens’ values.  It wasn’t obvious that expert decisionmaking was “better” than public decisionmaking on risks originating in publicly contested worldviews. Nor was it obvious that conflicts originating in conflicting worldviews could not be resolved by democratic decisionmaking procedures aimed at helping culturally diverse citizens to arrive at shared perceptions of the best available evidence on the dangers that society faces.

In his (very gracious, very intelligent) reply, Cass asserted that cultural cognition could simply be assimilated to his account of the reasoning deficits that distort public decisionmaking: “I argue,” he wrote “that insofar as it produces factual judgments, ‘cultural cognition’ is largely a result of bounded rationality, not an alternative to it.”  “[W]hile it is undemocratic for officials to neglect people’s values, it is hardly undemocratic for them to ignore people’s errors of fact” (Sunstein 2006)

This position—that cultural cognition and affiliated forms of motivated reasoning are rooted in “bounded rationality”—is now the orthodox view in decision science (e.g., Lodge & Taber 2013).

But we weren’t sure it was right.  As plausible as the claim seemed to be, it hadn’t been empirically tested.  So we set out to determine, empirically, whether the forms of information processing that are characteristic of cultural cognition really are properly attributed to overreliance on heuristic reasoning.

2.  A ten-year research program. The answer we arrived at over a course of a decade of research was that cultural cognition is not appropriately attributed to overreliance on the form of heuristic information processing associated with “System 1” reasoning.  On the contrary, the individuals in whom cultural cognition exerts the strongest effects were those most disposed to use conscious, effortful, “System 2” reasoning.

This conclusion was supported by two testing strategies.

The first was the use of observational or survey methods. In these studies we simply correlated various measures of System 1/System 2 reasoning dispositions with public perceptions of risk and related facts.

If public conflict over risk is a consequence of “bounded rationality,” then one should expect the individuals who evince the strongest disposition to use System 2 reasoning will form risk perceptions more consistent with expert ones than will individuals who evidence the strongest disposition to use System 1 forms of information processing.

In addition, one would expect polarization over contested risk to abate as individuals’ proficiency in System 2 reasoning dispositions increase: those individuals can be expected to “go with the evidence” and refrain from “going with their gut,” which is filled with heuristic-reasoning crap like “what do other people like me think?”

But in fact, those predictions are not borne out by the evidence.

In multiple studies, we found that the individuals who scored highest on one or another measure of the disposition to use conscious, effortful “System 2” information processing were in fact the most polarized on contentious risk issues, including the reality of climate change, the hazards of fracking, the danger of allowing citizens to carry concealed handguns etc. (Kahan, Peters et al. 2012; Kahan 2015; Kahan & Corbin 2016).

Inconsistent with the “bounded rationality” conception, this consistent finding is more consistent with the “cultural cognition thesis,” which posits that individuals can be expected to form identity-protective beliefs and to use all of the cognitive resources at their disposal to do so.

But to nail this inference down, we also conducted a series of experiments, the second type of testing strategy by which we probed Sunstein’s and others’ “bounded rationality” conception of cultural cognition and cognate forms of motivated reasoning.

These experiments consistently showed that individuals highest in the critical reasoning dispositions associated with System 2 information processing were using their cognitive proficiencies to ferret out evidence consistent with their cultural or ideological predispositions and to rationalize the peremptory dismissal of evidence inconsistent with the same (e.g., Kahan 2013).

Motivated Numeracy and Enlightened Self-government (Kahan, Peters et al. in press) reports the results of one of those studies.

3.  So what’s the upshot?  The original debate—over whether cultural cognition is a consequence of overreliance on System 1 heuristic processing—has been resolved, in my opinion.  Insofar as the individuals who demonstrate the greatest disposition to use System 2 reasoning are also the ones who most strongly evince cultural cognition, we can confident that it is not a “cognitive bias.”

But is it a socially desirable form of information processing on socially contested risks?

That’s a different question, one my own answer to which has been very much reshaped by the course of the “Ten Year Debate.”

It is in fact perfectly rational at the individual level to engage information societal risks in an identity-protective rather than a truth-convergent manner.  What an individual personally believes about climate change, e.g., won’t affect the risk she or anyone she cares about faces; whether as consumer, voter, public discussant, etc. her personal behavior will be too inconsequential to matter.

But given what positions on climate change and other societal risk issues have come to signify about who she is and whose side she is on in a perpetual struggle for status among competing cultural groups, a person who forms a position out of line with her cultural peers risks estrangement from the people on whom she depends on for emotional and material support.

One doesn’t have to be a science whiz to get this.  But if one is endowed with the capacity to make sense of evidence in the manner that is associated with System 2 information processing, it is predictable that she will use those cognitive resources to achieve the everyday personal advantages associated with the congruence between her beliefs and those of her cultural peers.

Of course, if everyone does this all at once, we are indeed screwed.  In that situation, diverse citizens and their democratically accountable representatives won’t converge, or converge nearly as quickly as they should, on the best evidence on the risks they genuinely face.

But sadly, this fact won’t change the psychic incentives that individuals have to use the forms of reasoning that most reliably connect their beliefs to the positions that signify membership in and loyalty to the identity-defining groups of which she is a member.

This is the tragedy of the science communications commons.

We should do something to dispel this condition.  But what?

That’s a hard question.  But it’s one for which an answer won’t be forthcoming if we rely on accounts of public risk perceptions that attempt to assimilate cultural cognition into the “public uses system 1, experts system 2” framework.

I suspect Cass Sunstein by this point would largely agree with everything I’m saying.

Or at least I hope he does, for the project to overcome “the tragedy of the science communications commons” is one that demands the fierce attention of the very best scholars of public risk perception and science communication.


Kahan. DM and Corbin, JC (2016) A Note on the Perverse Effects of Actively Open-minded Thinking on Climate Change Polarization. Research & Politics, 10.1177/2053168016676705.

Kahan, D.M. Climate-Science Communication and the Measurement Problem. Advances in Political Psychology 36, 1-43 (2015).

Kahan, D.M. Ideology, Motivated Reasoning, and Cognitive Reflection. Judgment and Decision Making 8, 407-424 (2013).

Kahan, D.M., Peters, E., Dawson, E. & Slovic, P. Motivated Numeracy and Enlightened Self Government. Behavioural Policy (in press).

Kahan, D.M., Peters, E., Wittlin, M., Slovic, P., Ouellette, L.L., Braman, D. & Mandel, G. The polarizing impact of science literacy and numeracy on perceived climate change risks. Nature Clim. Change 2, 732-735 (2012).

Kahan, D.M., Slovic, P., Braman, D. & Gastil, J. Fear of Democracy: A Cultural Evaluation of Sunstein on Risk. Harvard Law Review 119, 1071-1109 (2006).

Lodge, M. & Taber, C.S. The rationalizing voter (Cambridge University Press, Cambridge ; New York, 2013).

Sunstein, C.R. Laws of fear : beyond the precautionary principle (Cambridge University Press, Cambridge, UK ; New York, 2005).

Sunstein, C.R. Misfearing: A reply. Harvard Law Review 119, 1110-1125 (2006).

Leave a Comment