Cognitive biases in how we think about politics

Politics is a particularly contested domain when it comes to our processes of cognition. Political opinions can define our personalities, our social status, our familial relations, our employment, and sometimes our freedom. Those from the political science domain have gotten as far as agreeing that we do not act in a classically rational fashion, but have yet to cohere around an alternative model to the Simon’s (1955) and Down’s (1957) rational choice models. Over the last two decades some such as Lakoff (2012), Heidt (2001), Cohen (2003) and others have been working at attempting to define the parameters which have an effect upon our cognition, whether through affect, rationality, social cognition, or embodied cognition. In discussing this topic, I’ve considered the following papers:

Personality and Political Attitudes: Relationships across Issue Domains and Political Contexts – Gerber et al. 2010

Ideology, Motivated Reasoning, and Cognitive Reflection

Kahan 2012

Politics on the brain: An fMRI investigation

Knutson et al. 2006

Each one looks at how different aspects of ourselves intrude upon our political thinking.

Personality and Political Attitudes: Relationships across Issue Domains and Political Contexts – Gerber et al. 2010

The authors used a survey of registered voters conducted in the United States during the 2008 presidential election campaign in order to “examine the relationships between dispositional personality traits (the Big Five) and political attitudes”. The dispositional personality traits are part of McCrae and Costa’s five factor model (FFM) of personality. According to Gerber et al.’s perspective political opinions and attitudes are features of the characteristic adaptations of the individual. These personality traits (Basic Tendencies in figure 1 below) are the consequence of empirical findings that “individuals can be characterised in terms of relatively enduring patterns of thoughts, feelings and actions [and] that these traits can be quantitatively assessed”. The proceed from four assumptions about human nature “knowability, rationality, variability, and proactivity” (McCrae and Costa 1999)

Knowability is the assumption that it is possible to scientifically study human personality.

Rationality is the assumption that people are generally capable of understanding themselves and others.

Variability is the assertion that we differ from each other psychologically in significant ways

Proactivity is the notion that people are not merely passive victims of circumstances (though some may be) but rather that their life is a reaction of their personality to these circumstances.

The key feature of this model is “the distinction between basic tendencies… and characteristic adaptations” where characteristic adaptations are the “the concrete manifestations [of basic tendencies] in the personality system” (McCrae and Costa 1999 p162) according to this model, while characteristic tendencies vary according to life circumstances, basic tendencies do not

Figure 1, Pervin & John (1999)

Through this lens Gerber et al. hold that as culturally conditioned phenomena, and incorporating as they do aspects of Self-Concept, political attitudes may be defined as characteristic adaptations, part of the reaction of the biologically constrained personality traits and the environment the individual finds themselves part of.

A concern with respect to this theoretical model is the limitation that it is an almost purely empirical model. It describes certain facts, that individuals have things which can be measured the five tests that make up the FFM, and that these measurements are stable over long timeframes, and that these traits result in behaviours that are consistent over wide swaths of the population.

It doesn’t offer us proposals for the mechanisms which cause these traits to impact upon behaviours. It does not offer causal explanations for how these traits emerge and stabilise in individuals. It is a taxonomic model of categorises, rather than one which offers us insight, this would be ok if this model was in fact stable over time, however Soto et al. (2010) have developed findings that suggest that traits can vary with the age of the participant, and that within traits there can be sub-traits which exhibit relatively large degrees of variance compared with the other components of that trait as the person ages.

This suggests that age would have to be carefully considered within a model that assumed that these were stable variables.

Another issue is that the procedure for this study was to carry out an internet poll which used “sampling and matching techniques” to account for selection bias (towards those interested in politics) when conducting polls such as these.

Finally this is a fishing expedition, in so far as there is a general sense that these, and other socio-cultural factors may have an effect upon the voting public, however assuming that they were careful to manage the selection bias it is a survey which may turn up unexpected results, however given that there are a lot of variables used, and the authors conducted OLS, logit, and also the tobit regressions that are shown in the paper there does appear to be the risk of false positives.

The most robust finding was the discrepancy between black and white participants of the survey. There was a wider variance among black voters and white voters, though it is not possible to tell if this is a consequence of the black population being a much smaller sample than their white peers. There is also consistent discrepancies between the means for these two populations, with black people seeming to be more centrist than white people.

If the cultural effect is robust, this could prove to be an interesting area for conducting further research, there is some evidence from the Irish context that there exists a cultural cleavage between different sets of voters (Byrne O’Malley 2012) but most research in the area assumes relatively simple relationships between demographics and politics, unmediated by personality traits and complex interactions with personality traits.

Ideology, Motivated Reasoning, and Cognitive Reflection

Kahan 2012

Kahan compares three different theoretical models in assessing the impact of motivated reasoning on the participants of this study. Two of these models, Bounded-Rationality Thesis (BRT) and Neo-Authoritarian Personality Thesis (NPT) are “System 1”/“Hot”/“Heuristic” models of thinking, presuming quick actioned thoughts which are flawed in terms of how they accept and process information which conflicts with their pre-existing biases.

BRT, Kahan argues, is a function of skewed beliefs regarding risks. As different risks are attributed to different risk sources, or policies, then Kahan expected “to see systematic differences in risk perceptions across members of ideologically or culturally uniform groups”.

NPT in contrast is a theory that holds that politically conservative people engage in motivated thinking as a consequence of their “low quality information processing abilities” which would require them to rely on the easier System 1 cognitive mechanisms that the more challenging System 2 ones. Furthermore, this theory holds that the alignment of the “Openness” personality trait with liberal ideology should skew those with a right-wing ideology towards system 1 motivated thinking (the assumption being that closed minded people would have to engage in deeper levels of motivated thinking that people who are ostensibly open to new ideas).

BRT in contrast would not expect that there would be a schism along ideological lines when it came to the use of System 1 processing.

The third theory examined by Kahan is the Expressive Rationality Thesis (ERT) and it supposes that for the individual “whatever she as a single individual does—as consumer, as voter, as participant in public discourse—will be too inconsequential to have an impact”. This would suggest that in her internal weighing of which side of the argument to come down upon the actual merits or demerits of that argument is irrelevant, what matters is what others think of her opinion.

This perspective sees political beliefs and opinions as signifiers of group affiliation not things which have an external validity of their own, thus the question she asks herself is not the classically rational question, “Where does the truth lie?” it is instead, “What do people like me think about this kind of thing?”. This can decompose into a tragedy of the commons type scenario where a group can collectively get trapped in a sub-optimal Nash equilibrium as a consequence of everyone holding the same wrong opinion where for any individual to take any other option would see them worse off an the collective no better off.

If the challenge is to cohere with the group position, Kahan argues that those who have greater facility with System 2 processes

Under this hypothesis, reasoning will not lead one towards truth, but rather the kind of thing that will not damage the rationaliser’s standing in their “ideologically or culturally defined affinity groups”. Kahan argues that with many political, social, and moral issues, from climate change to gun control, an individual may personably be unable to do anything to ameliorate the problem, but the cost of stepping outside of the consensus of their social group can be huge “as opposing positions on these issues have come to express membership in and loyalty to opposing self-defining groups, a person’s formation of a belief out of keeping with the one that predominates in hers could mark her as untrustworthy or stupid, and thus compromise her relationships with others”.

In order to find a solution to the quandary posed by studies that show people are willing to ignore evidence contrary to their opinion, or even form to more hard line opinions in spite of the contrary information (the “blowback effect”) Kahan outlines how it is expressively rational “for individuals in [a] situation [where they are likely to pay a social price for expressing a particular opinion] to assess information in a manner that aligns their beliefs with those that predominate in their group” even though the group as a whole may be engaging in a suboptimal behaviour.

This is coherent with the Taber et al. (2009) finding “that individuals who are more politically sophisticated—ones who are more interested in and know more about politics—reliably avail themselves of their greater political knowledge when affectively motivated to defend ideologically congruent beliefs” (Kahan 2012).

Contrasting ERT this with other theories, BRT and NPT, the expectation is that there would not be an ideological skewing of System 1 decision making which NPT predicts, and that the reliable employment of more effortful, conscious information processing will magnify the polarizing effects of identity-protective cognition. Furthermore this type of skewing in favour of group affiliation should include other groups (university membership, or being a sports fan) which are not in themselves ideologically grounded.

The critical variation between these theories is whether we carry out our political thinking through affect processing or through rational cognitive mechanisms, the alternative perspective is that it is not this simple, there may be a complex combination of processes some of which may be operating in classically rational ways others while others may be processed emotionally. Furthermore not all thinking is verbal, autobiographical memories can create frames which can affect even the most rational of people, in other situations it is possible to solve some problems using visual processing, and still other ways of thinking are effected through embodied cognition. In these non-verbal contexts it’s not certain that the hot/slow, system 1/system 2 dichotomies have any meaning. Is there a rational way to kick a ball for example?

With framing this study in an either/or fashion there’s the risk that other aspects of thought are thrown out along with the theories that are ostensibly falsified.

The experiment surveyed a fairly representative (if slightly over educated) sample of the US population. Having gathered the demographic data the survey conducted a Cognitive Reflection Test (CRT) to test the cognitively flexibility of the participants. It then streamed the participants into three arms. One was a neutral control statement regarding global warming, the second was a statement biased in favour of global warming believers, and the third was a statement biased in favour of global warming deniers, what this test seems to show is that those who did better at the intelligence test, and were more ideologically minded were more biased in favour of their group of ideological alignment.

What this experiment established was that there is an effect between being mathematically adroit and using those abilities to determine whether your arm of the assessment was undermining or supporting your fellow ideological group members. This result is coherent with some other work in the area and neatly falsifies the NPT hypothesis (that right-wingers are more stupid than left-wingers). It also suggests that the BRT is at least over-simplified, which probably isn’t a bad thing, if the answer about a complicated topic like cognition is always hot irrational processing, we’re probably asking the wrong questions.

Politics on the brain: An fMRI investigation

Knutson et al. 2006

Knutson et al. take a functionalist view of the brain. Their participants filled in a questionnaire outlining their political opinions and their political affiliation and the parties their parents voted for. The participants also carried out an Implicit Attitudes Test (IAT) inside of an fMRI. “The IAT measures strengths of associations between concepts by comparing response times on two combined discrimination tasks. Participants are required to sort stimuli representing four concepts using just two responses, each assigned to two of the four concepts” (Egloff and Schmukle 2002). The theory that rests behind the IAT is emotional-cognitive dissonance. The participants are expected to like one of each type of sort stimuli, and find it more difficult (slower or more error prone) to correctly sort nice stimuli with negative stimuli, while finding it easy to sort nice with nice, and negative with negative. Typically this is conducted with images of black people and white people, contrasted with nice words e.g. “Sweet” and negative words “Rotten”, where people find it easy to sort negative words with black people and easy to sort nice words with white people they are presumed to have a bias towards black people. Knutson et al. also reference Damasio’s somatic marker hypothesis (1994) which holds that we have brain structures that function as emotional processing centres. Having observed patients with damage to the ventromedial prefrontal cortex (VMPFC), Damasio argued that the VMPFC is where linkages between “the facts that compose a particular situation, and the emotion previously paired with it in an individual’s contingent experience”.

Inside the fMRI, the participants conducted an IAT which contrasted faces of politicians who were equally weighted in terms of republican and democrats.

An assumption that researchers made was that they would be able to detect activity in the VMPFC linked with affect, accompanied perhaps by an amygdala response, where there was an out-group politician, and that activity in the dorsolateral prefrontal cortex (DLPFC). This was only the second occasion that someone conducted an IAT in an fMRI. The previous occasion contrasted flowers with insects, so this was the first time someone conducted an IAT in an fMRI with human faces. The experimenters conducted their test with a novel set of images, and also interspersed names of politicians with the photos of the politicians. They also had the participants rate the politicians on how likely they were to vote for them, and also regarding how high profile the participants thought the politicians to be. Before finally smoothing out the fMRI data into 8mm³ voxels.

I suspect that they may have been overly ambitious in their aims.

The theoretical commitment of the researchers was a kind of naive functionalism. They clearly thought that there would be a relatively simple mapping along the affect and cognitive processing systems between the two political parties. They hadn’t even checked to see if their IAT (with additions) was robust before getting participants to use the fMRI.

What they discovered was that the names of politicians had no significant effect, meanwhile the participants rated their likelihood to vote for a given candidate in a manner that did not align with what they said their political affiliations were. The researchers then had to use the responses of the participant in the machine to determine what their ‘real’ affiliation was, which suggests that there may be problems with the IAT “D score” which is the only positive result of the experiment.

It is rarely clear what an fMRI is showing, and it is hard to determine whether it is capable of returning a valuable answer under the circumstances of this experiment. The researchers didn’t seem to have an alternative hypothesis.

The experimenters should have started from somewhere much simpler. They could have conducted a White/Black or Male/Female IAT within an fMRI which would have had a body of work to contrast against, and allow there to be a data set from which one could begin to interpret fMRI results with respect to cognitive biases. They could have tests their experimental materials before the full study which would have allowed them to establish the existence of the ‘false report’ regarding political affiliation. They could also then have tested for an fMRI effect, in case doing the test in an fMRI had an effect on the participants.

Conclusion

From results such as Gerber et al. there are suggestions that there are new finding to be made within this domain, but they are likely to be difficult to find and subtle in their effect. Kahan has shown how a good tight experimental design built around conflicting theories can elicit robust information, the risk of his particular approach is that it may leave too wide a territory to the excluded middle, especially when we see multipolar dimensionality arising from results such as Gerber et al.’s. Finally there is Knutson who shows the risk of being a hedgehog and knowing only one big thing. The failure to develop a robust result was not just that there was a poorly designed experiment, but that there was an opportunity to gather real information with valuable resources that was missed.
Bibliography

Byrne, Kevin P., and Eoin O’Malley. “Politics with hidden bases: unearthing the deep roots of party systems.” The British Journal of Politics & International Relations 14.4 (2012): 613-629.

Cohen, G. L. (2003). Party over policy: The dominating impact of group influence on political beliefs. Journal of personality and social psychology, 85(5), 808.

Damasio, A. R., Everitt, B. J., & Bishop, D. (1996). The somatic marker hypothesis and the possible functions of the prefrontal cortex [and discussion]. Philosophical Transactions of the Royal Society B: Biological Sciences, 351(1346), 1413-1420.

Downs, Anthony. “An economic theory of political action in a democracy.” The journal of political economy (1957): 135-150

Egloff, B., & Schmukle, S. C. (2002). Predictive validity of an implicit association test for assessing anxiety. Journal of personality and social psychology, 83(6), 1441.

Gerber, A. S., Huber, G. A., Doherty, D., Dowling, C. M., & Ha, S. E. (2010). Personality and political attitudes: Relationships across issue domains and political contexts. American Political Science Review, 104(01), 111-133.

Haidt, J. (2001). The emotional dog and its rational tail: a social intuitionist approach to moral judgment. Psychological review, 108(4), 814.

Kahan, D. M. (2012). Ideology, motivated reasoning, and cognitive reflection: an experimental study. Judgment and Decision Making, 8, 407-24.

Knutson, K. M., Wood, J. N., Spampinato, M. V., & Grafman, J. (2006). Politics on the brain: An fMRI investigation. Social Neuroscience,, 1(1), 25-40.

Lakoff, G., & Wehling, E. (2012). The little blue book: The essential guide to thinking and talking democratic. Simon and Schuster.

Pervin, L. A., & John, O. P. (1999). Handbook of personality: Theory and research. Elsevier.

McCrae, R. R., & Costa Jr, P. T. (1999). A five-factor theory of personality. Handbook of personality: Theory and research, 2, 139-153.

Simon, Herbert A. “A behavioral model of rational choice.” The quarterly journal of economics (1955): 99-118

Soto, C. J., John, O. P., Gosling, S. D., & Potter, J. (2011). Age differences in personality traits from 10 to 65: Big Five domains and facets in a large cross-sectional sample. Journal of personality and social psychology, 100(2), 330.