Institut Jean Nicod

Accueil du site > Séminaires/Colloques > Archives > Séminaires > 2019-2020 > Collective intelligence > Collective intelligence

Collective intelligence


Session 1

Sami Yousif, Yale University 
4pm, Wednesday May 6th 

The illusion of consensus : How adults and children evaluate a ’true’ and ’false’ consensus

When evaluating information, we cannot always rely on what has been presented as truth : Different sources might disagree with each other, and sometimes there may be no underlying truth. Accordingly, we must use other cues to evaluate information—perhaps the most salient of which is consensus. But what counts as consensus ? Do we attend only to surface-level indications of consensus, or do we also probe deeper and consider why sources agree ? I’ll present four experiments that demonstrate how individuals evaluate consensus. Strikingly, people seem to evaluate consensus only superficially : Participants were equally confident in conclusions drawn from a true consensus (derived from independent primary sources) and a false consensus (derived from only one primary source) — and this is true even immediately after participants explicitly stated that a true consensus was more believable than a false consensus. I’ll also show some work with children tracking how they come to evaluate consensus over time. Finally, I’ll discuss how this illusion of consensus acts as a powerful means by which misinformation may spread. 



Session 2

Alexandre Noyes, Yale Dept. of Psychology
4pm, Wednesday May 13th 

The flexibility of kind representations

A central question in the cognitive sciences is the nature of concepts ; in particular, the nature of concepts of meaningful and richly structured categories like tigers and gold (what I call “kinds”). Psychological essentialism proposes that children and adults represent kinds as possessing internal, naturalistic causes like genes and chemical make-up. However, children and adults also acknowledge kinds that are socially constructed : in particular, institutional kinds like money, academic departments, and lawyers. In sharp contrast to essentialism, participants reason that these entities exist because individuals continually recognize and enact their existence. Despite their unique features, socially constructed categories conform to general principles of kinds : e.g., They license generic statements, they have a reality distinct from their superficial appearances, and they support robust predictions and explanations. Consequently, they demand revision of current theory. Psychological essentialism fails to capture the breadth of ordinary categories : Participants reason that many kinds are entirely socially constructed rather than essential.



Session 3

Hugo Mercier, IJN/CNRS/ENS-Ulm
4pm, Wednesday May 20th 

Not born Yesterday : Why humans are less gullible than we think.

It is often thought that humans are gullible, easily manipulated by demagogues, advertisers, and politicians. I will argue that the opposite is true : humans are equipped with a set of psychological mechanisms that allow them to properly evaluate communicated information, and to reject information that is false or harmful. I will rely on experimental psychology data, as well as studies showing the failures of mass persuasion, from Nazi propaganda to American presidential campaigns. I will also offer explanations for the success of some misconceptions—from pizzagate to flat earth—that are not based on credulity. 



Session 4

Brent Strickland, IJN/CNRS/ENS-Ulm
4pm, Wednesday May 27th 

“Myside bias” in the production, consumption, and dissemination of science

Science produces reliable information that can have a direct impact on our daily lives. The current COVID-19 crisis provides some particularly clear examples of this general observation. The machinery of science is our best hope for developing and evaluating vaccines and treatments. And technical experts are supplying important evidence based advice shaping crisis response strategies (e.g. confinement policies informed by epidemiology). For all the positives however, it is also no secret that scientific practice and how we use its insights are far from perfect, and sometimes deeply flawed. The current talk examines one particularly salient source of systematic error : The “myside bias.” This term refers to a deep seated and universal human tendency to preferentially seek out and evaluate information which supports one’s desires or pre-existing beliefs. Across a range of behavioral experiments and computational models I show how the myside bias systematically influences the production, consumption and dissemination of scientific claims. One of the surprising lessons that comes out of this work is that bias is not always bad, but can also produce benefits when considered at an organizational or societal scale. For example, while groups with a higher degree of myside bias are more error prone (clearly bad), they are also faster to make correct decisions. A better understanding of these sorts of tradeoffs associated with bias can help to better anticipate and evaluate the likely downstream consequences of changes to incentive structures or procedures shaping scientific communities, and can contribute to improved public interaction with the scientific community.



Session 5

Emory Richardson, Yale Dept. of Psychology
4pm, Wednesday June 3rd 

Does informational independence always matter ? Children believe small group discussion is more accurate than ten times as many independent informants

Learners faced with competing statements that each have support from multiple sources must decide whom to trust. Lacking firsthand knowledge, they frequently trust the majority. Yet, majorities can be misleading if most members are relying on hearsay from just a few members with firsthand knowledge. Thus, past work has emphasized the importance of informational independence when deciding whom to trust, showing that children and adults do consider informational independence important in certain contexts. However, because informational independence precludes group deliberation, we ask whether children make the reverse inference and devalue informational independence when facing a problem that could benefit from deliberation. In two studies, children and adults ignore informational independence when attempting to answer abstract reasoning questions. However, for a question type for which deliberative reasoning would be of doubtful benefit, children and adults seek advice from multiple independent sources rather than a deliberative group.



Session 6

Hannah Hok, Univ. of Chicago, Dept. of Psych.
4pm, Wednesday June17th 

Politics on the playground : children’s intuitions about moral condemnation and majority rules voting


Children inhabit a world rich with sophisticated social information and interaction, which requires equally rich and sophisticated social intuitions. Here, I present two lines of research investigating children’s early social reasoning. The first set of studies examine what inferences children make about others based off of their moral condemnation. Across four studies, we find that by the age of 7-years-old children treat condemnation as a signal of one’s moral commitments—thinking that a condemner is less likely to engage in the immoral behavior and that she should be punished more harshly if she is caught hypocritically engaging in the action they condemned. In a second set of studies, we examine children’s intuitions about majority rules voting as a way groups make decisions. Across two studies, we demonstrate that children as young as 4-years old think one should use majority rules rather than minority preference to make group decisions. Further, by age 6-years-old, children think that one should resolve group decisions by majority rules voting rather than another impartial decision-making procedure (here coin flip). Importantly, we demonstrate that this is not because children think that majority rules voting is always fair ; they understand that voting is unfair when it leads to the “tyranny of the majority”. 



Session 7

Sacha Yesilaltay, IJN/CNRS/ENS-Ulm
4pm, Wednesday July 1st 

Fake news and misinformation


Fake news created with the intention of generating engagement is not constrained by reality. This freedom allows fake news to tap into the natural biases of the human. But despite the attractiveness of fake news stories, most people are reluctant to share them. Why ? I will present evidence that sharing fake news hurt one’s reputation in a way that is difficult to fix, and that these reputational costs partly explain why most people are reluctant to share fake news. However, some people share fake news stories. And interestingly some of them share (fake) news they suspect to be inaccurate. Why would they do that ? I will present one factor, that alongside accuracy, drives the sharing of true and fake news : the ‘interestingness-if-true’ of a piece of news. I will argue that people may not share news of questionable accuracy by mistake, but instead because the news has qualities that make up for its potential inaccuracy, such as being interesting-if- true.



Session 8

Antoine Marie, ENS-Ulm/UM6P SCI
4pm, Wednesday July 8th 

The psychological consequences of moral conviction


From a functional standpoint, one of the primary functions of moral processes is to motivate individuals to pursue (what their brains perceive as) sources of cooperation benefits and to avoid social costs. A feat that is typically achieved by a psychology of moral conviction or moralization according to which certain behaviours ought or ought not be performed, and certain moral causes ought to be pursued or combatted, in ways that can be relatively insensitive to side effects. For instance, strongly committed anti-nuclear activists may want to politically prioritise renewable energies while being somewhat blinded to their low efficiency in combatting climate change compared to nuclear fission.

Across a series of online experiments conducted with Brent Strickland, Sacha Altay and others, I explore how variations in the degree to which participants moralize a political issue biases their moral judgments and behaviors. 

Project 1 finds that, when judging a powerful actor’s actions (CEO or Minister), greater moral conviction on a political value (e.g. protecting the environment, regulating immigration) predicts greater weight given to the actor’s ‘good intentions’ to honour the value, but doesn’t consistently predict greater valuation of his decision’s being effective in actually promoting the value on a practical level. Those findings suggest that politicians may be mostly judged on the good virtues they express rather than on the basis of their actual competence. Project 2 shows that greater moral conviction on a political issue (e.g. gun control, abortion) increases people’s tendency to pass along ideologically congruent (vs. incongruent) news headlines, whether true of false, that touch on the issue. This transmission bias, which can contribute to maintain starkly different priors across political subcultures (e.g. US Republicans and Democrats), is not easily mitigated by corrective interventions.