Soutenance de thèse - Antoine Marie - "Moral conviction, political polarisation and susceptibility to misinformation""

La soutenance aura lieu jeudi 25 mars 2021, de 14h à 18h

Le jury sera composé de :

Jean BARATGIN (Université Paris 8) directeur de thèse
Gérald BRONNER (Université de Paris) membre invité
Sabine GUERAUD (Université Paris 8)
Hugo MERCIER (CNRS) rapporteur
Cathal O’MADAGAIN (Université Mohammed V1 Polytechnique) rapporteur
Gloria ORIGGI (CNRS) examinatrice
Brent STRICKLAND (CNRS) examinateur

Titre de la thèse  : Moral conviction, political polarisation and susceptibility to misinformation

Résumé : 

In the political arena, rational decision making should be based on accurate assessments of the costs and benefits associated with technologies and policies—from GMO foods to nuclear power to global warming and gun control—lest immense opportunities may be missed.

However, human minds are home to many cognitive biases, many of which likely evolved by natural selection. Those biases distort information processing and transmission, standing in the way of people indexing their beliefs on the scientific consensus when it exists, and maintaining nuanced representations of complex issues when the evidence is equivocal. This thesis explores various cognitive biases contributing to this outcome through one theoretical and two empirical papers.

“The cognitive foundations of misinformation on science” (Chapter 2) draws on extant research in cultural evolution, science communication, and evolutionary psychology to review some of the main psychological mechanisms involved in causing widespread misbeliefs on policy-relevant scientific and technological issues. Among other culprits, the article highlights the reconstructive character of human communication and the role played by expectations of relevance in exaggerating factual claims ; the myside bias and our tendency to moralize and politicize factual topics ; our preference for homophily ; our susceptibility to threatening information and conspiratorial thinking ; and the illusion of understanding.

“Moral conviction predicts sharing preference for politically congruent headlines” (Chapter 3) explored, through 8 experiments, what type of political information US participants decide to share on simulated social media on controversial topics such as gun control or abortion. In analogy with our myside bias when processing information, we found that participants have a sharing preference for politically congruent news stories which increased with the moral importance of the issue—whether the news was true or false. We also found that this sharing preference was little swayed by manipulations of the audience composition anonymousness of the account from which sharing is done, and exposition to an intervention message warning against political bias. Perceived accuracy and usefulness for one’s political goals were among the main predictors of sharing. We suspect that a tendency to do selective communication may accelerate discrepancies in prior beliefs across political subcultures, undermining the ability of liberals and conservatives’ partisans to find common ground on important policy issues.

Across 5 experiments run on French subjects, “Intentions matter a lot, and efficiency little, in folk judgments of policy decisions” (Chapter 4) found that laypeople spontaneously prefer altruistically motivated policies that achieve little at a high cost over selfishly motivated but highly efficient and profitable policies. This preference appeared to be driven by a combination of low responsiveness to differences in efficiency expressed in numeric format—a tendency magnified by a participant’s degree of moral conviction on the issue— and of high sensitivity to the intentions driving the person implementing the policy.

Taken together, this work contributes to portray humans as a moralistic and paranoid species, wired for the righteous defense of political causes and moral principles more than for the pursuit of what may effectively promote collective welfare.