Post last updated 4th Jan 2019
We recently published a paper in Current Biology entitled “Metacognitive Failure as a Feature of Those Holding Radical Beliefs”, which identified a link between radical world views and metacognition. Unsurprisingly this got picked up in a few media outlets. In a recent post on Twitter, Prof. Ken Miller at Columbia University queried the interpretation of the findings:
We think that these comments stem from a misunderstanding about the relationship between radicalism and extremism, which we’re glad of the opportunity to clarify.
Ken says: “The claim is that radicals, which you would think would mean the extremes of political left and right…”
This is incorrect, and the source of the subsequent confusion. The paper did not set out to examine extremism, and we specifically avoided using this term in both the paper and the UCL press release. Instead our focus was on an index of radical views closely modelled on previous measures in the literature. As we say in the paper:
These questionnaires were selected based on prior models of political radicalism as stemming from a combination of intolerance to others’ viewpoints, dogmatic and rigid beliefs, and authoritarianism, which represents adherence to in-group authorities and conventions, and aggression in relation to deviance from these norms [23–25].
Some (but not all) of the press coverage (mainly the headlines) extrapolated from our findings on radicalism to write about political extremism. But we don’t think our measure is specific to politics:
…we stress that radicalism is likely to reflect a general cognitive style that transcends the political domain—as exemplified by links between religious fundamentalism and increased dogmatism and authoritarianism [22, 26]—and instead refers to how one’s beliefs are held and acted upon .
Similarly, while our measures of radicalism were derived from questionnaires tapping into political attitudes, it is possible that impairments in metacognition may constitute a general feature of radicalism about political, religious, and scientific issues.
Our key result is that a radicalism index (formed from the combined measures of dogmatism and authoritarianism) is negatively related to metacognitive sensitivity (the ability to distinguish correct from incorrect decisions in the perceptual task), but not task performance. These effect sizes are indeed small, but robust and replicable in a second independent dataset. Given that the tasks are far removed from real-world issues we think it’s striking that basic difference in metacognition predict answers to questions indicative of radical beliefs:
Despite relatively small effect sizes, our findings linking radicalism to changes in metacognition are robust and replicable across two independent samples. However, we note that other, domain-specific facets of metacognition (e.g., insight into the validity of higher-level reasoning or certainty about value-based choices ) are arguably closer to the drivers of radicalization of political and religious beliefs, suggesting that the current results represent a lower bound for the strength of a relationship between metacognitive abilities and radicalism.
While our main focus was on radicalism, we also examined relationships with political orientation. First, we found that conservatism was related to overall confidence, but not metacognitive sensitivity (see Figure 3C in the paper). Again, this result shows that radicalism is not the same as being on the left or right of the political spectrum; instead, these two aspects of the questionnaire data relate to different facets of task performance.
In an earlier version of the paper, we also examined whether these facets of radicalism also mediated the extremity (absolute value) of political orientation. We’re glad of the opportunity to revisit it here. A multiple mediation model relating metacognition, facets of radicalism and the political extremity is shown below (data are pooled across Studies 2 and 3):
Model estimated with lavaan in R; all parameters are standardized; *P<0.05, **P<0.005, ***P<0.001
Importantly, political extremity in isolation is not linked to metacognitive sensitivity; however, the relationship between metacognition and political extremism is mediated by its impact on dogmatism and authoritarianism. This is consistent with the findings of a recent paper by Leor Zmigrod and colleagues at Cambridge, who showed that the impact of psychological flexibility on political attitudes (towards Brexit) was mediated by its effect on ideology.
So, to summarise, we find that radicalism but not extremism is linked to metacognitive failure. There is a complex relationship between these constructs (as Ken points out in his post), and causality remains to be disentangled. But we think it’s plausible that the differences in metacognitive function we find in relation to radical views (the focus of the paper) may in turn predict the extremity of political orientation.