Working memory and need for cognition influence beliefs in conspiracy theories

ABSTRACT

We raise the hypotheses that there is a main effect of Working Memory (WM) as well as an interaction between WM and Need for Cognition (NC) on subjects’ propensity towards believing in conspiracy theories. An empirical study confirms that subjects with high WM capacities tend to hold conspiratorial beliefs to a lesser extent than subjects with lower WM capacities. An interaction effect was also present as beliefs in conspiracy theories tended to increase with increasing levels of NC for subjects with low WM capacities but to decrease for subjects showing both high WM capacities and high NC levels. Reported results persist in the presence of paranoid ideations (a strong and reliable predictor of conspiratorial beliefs) as a control measure.

KEYWORDS:

• Working memory
• need for cognition
• conspiracy theories

Data availability

Data is available to all interested parties.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 The sample size was chosen prior to the study with the intent to achieve power of around 80% to detect medium-to-low effect sizes (Cohen, 1992).

2 Briefly, Lasso strives to minimize the model’s error while simultaneously penalising the absolute values of regression coefficients. As a result this procedure tends to produce a solution which sets some of the regression coefficients to 0. The degree of penalty, λ, can be set by the researcher but it is often chosen on the basis of a cross-validation procedure which selects the value of λ so as to minimize the average error obtained for the validation sub-samples. Here, we used a 10-fold cross-validation procedure which resulted in λ = 0.06 for the completely standardised (z-scores) dataset.

3 Swami et al. (2011) report a standardised regression coefficient of Crystalized Intelligence on BCTI equal to −0.11 while the standardised coefficient of WM on the same dependent measure (Table 2) is more than twice as large. Raw correlation coefficients follow the same pattern (−0.19 vs. −0.31). As our study is limited to subjects with higher education (note also that we have a somewhat restricted age range), it can be expected that the effects of WM on conspiratorial beliefs may be even larger among the general population (i.e. we expect to observe larger WM variance among the population at large and hence an increased covariance with beliefs in conspiracy theories).

4 In more technical terms, we are referring to an application of the availability heuristic (e.g. Tversky & Kahneman, 1973)