Odds Are
Overconfidently Conspiratorial: Conspiracy Believers are Dispositionally Overconfident and Massively Overestimate How Much Others Agree With Them
Gordon Pennycook, Jabin Binnendyk & David Rand
Personality and Social Psychology Bulletin, forthcoming
Abstract:
There is a pressing need to understand why people believe in conspiracies. Although past work has focused on needs and motivations, we propose an alternative driver of belief: overconfidence. Across eight studies with 4,181 U.S. adults, conspiracy believers consistently overestimated their performance on numeracy and perception tests (even after taking their actual performance into account). This relationship with overconfidence was robust in controlling for analytic thinking, the need for uniqueness, and narcissism, and it was strongest for the most fringe conspiracies. We also found that conspiracy believers -- particularly overconfident ones -- massively overestimated (>4×) how much others agree with them: Although conspiratorial claims were believed by a majority of participants only 12% of the time, believers thought themselves to be in the majority 93% of the time. This was evident even when asked to rate agreement among counter-partisans, indicating that conspiracists are genuinely unaware that their beliefs are on the fringe.
Algorithmic Personalization of Information Can Cause Inaccurate Generalization and Overconfidence
Giwon Bahg, Vladimir Sloutsky & Brandon Turner
Journal of Experimental Psychology: General, September 2025, Pages 2503-2522
Abstract:
Personalization algorithms are widely used online to deliver recommendations fine-tuned to individual users. This specificity comes at the cost of the diversity of information presented to users, limiting exposure to alternative perspectives and potentially reinforcing existing beliefs. We investigated the degree to which personalization can hinder the acquisition of new knowledge of categories. We asked participants to learn about alien categories under different levels of personalization and tested their knowledge using a postlearning categorization task. Our results show that learners in personalized environments sample feature information more selectively during the learning phase and develop inaccurate representations about the categories. Critically, they also report inflated confidence about their inaccurate decisions for categories for which they had little exposure. Our results suggest that personalization can distort learners’ understanding of the environment, bias information sampling, and induce incorrect generalization of knowledge.
Overconfidence Persists Despite Years of Accurate, Precise, Public, and Continuous Feedback: Two Studies of Tournament Chess Players
Patrick Heck et al.
Psychological Science, forthcoming
Abstract:
Overconfidence is thought to be a fundamental cognitive bias, but it is typically studied in environments where people lack accurate information about their abilities. We conducted a preregistered survey experiment and replication to learn whether overconfidence persists among tournament chess players who receive objective, precise, and public feedback about their skill. Our combined sample comprised 3,388 rated players aged 5 to 88 years from 22 countries with an average of 18.8 years of tournament experience. On average, participants asserted that their ability was 89 Elo rating points higher than their observed ratings indicated -- expecting to outscore an equally rated opponent by nearly 2 to 1. One year later, only 11.3% of overconfident players achieved their asserted ability rating. Low-rated players overestimated their skill the most, and top-rated players were calibrated. Patterns consistent with overconfidence emerged in every sociodemographic subgroup we studied. We conclude that overconfidence persists in tournament chess, a real-world information environment that should be inhospitable to it.
Repeating Statements Increases Source Credibility
Simone Mattavelli, Marco Brambilla & Christian Unkelbach
Personality and Social Psychology Bulletin, forthcoming
Abstract:
Repeating statements increases their perceived truth. Yet, whether repetition enhances the credibility of their source remains unexplored. We examined a repetition-induced source credibility effect in four preregistered experiments. In Experiment 1 (N = 90), we exposed participants to 20 unrepeated and 20 repeated statements communicated by 40 individuals. Repetition significantly increased both statement truth and source credibility. Experiment 2 (N = 65) tested if the increase in source credibility generalizes to new contexts. After rating the truth of repeated and unrepeated statements paired with different sources, participants judged novel statements from the same sources, showing increased credibility for sources previously associated with repeated statements. Experiment 3 (N = 180) did not replicate the effect without initial truth ratings. However, Experiment 4 (N = 435) resolved this inconsistency, showing a repetition-induced source credibility effect, regardless of participants’ task during source–statement pairings. We discuss the implications of these findings for illusory truth literature and persuasion and communication management.
The “Confidence” Trap: When the Likelihood Is Low, Forecasters Look Less Competent When They Refer to Their Confidence in an Outcome Rather than Its Probability
Mauricio Palmeira & Haipeng (Allan) Chen
Journal of Consumer Research, forthcoming
Abstract:
Marketers make various forecasts, including those about new products, financial instruments, sporting events, and medical procedures, to influence consumer decisions. In communicating the likelihood that a forecaster assigns to an outcome, the forecaster can refer to their confidence in the outcome (e.g., “I’m 30% confident”) or the probability of the outcome (e.g., “There is a 30% probability”). We propose that the choice of language (e.g., probability vs. confidence) affects the perceptions of forecasters, yielding predictable consequences on consumer decisions. Specifically, we argue that some languages (e.g., confident/sure/certain) encourage internal attributions (e.g., to a forecaster), whereas other languages (e.g., probability/likelihood/chance) encourage external attributions (e.g., to an outcome). As a result, expressions of a forecaster’s confidence (vs. outcome probability) make the forecaster look less competent, especially when the likelihood of the outcome is low. A series of studies shows the effect in various consumption scenarios. We further show that the effect is mediated by internal (vs. external) attributions, influences real betting decisions, and is mitigated when the likelihood is high and among consumers with a weaker tendency to make internal attributions (ie, a weaker correspondence bias).
An information-theoretic foreshadowing of mathematicians’ sudden insights
Shadab Tabatabaeian et al.
Proceedings of the National Academy of Sciences, 2 September 2025
Abstract:
The “eureka” insights that drive progress in science and mathematics remain shrouded in mystery. Sudden, unexpected, appearing like “flashes of lightning”, these insights have the hallmarks of critical transitions in complex systems. Here, zooming in on mathematicians working on proofs in their own departments, we show that sudden insights are anticipated by a system-agnostic, information-theoretic early warning signal. Using dense behavioral recordings of mathematicians’ moment-to-moment activity, we find that their blackboard interactions (e.g., writing, gesturing; N = 4,653) became increasingly unpredictable before an insight, analogous to the critical fluctuations that anticipate transitions in physical and ecological systems. We explore analytically when this early warning signal applies to varied systems with discrete, symbolic dynamics. While bibliometric analyses offer a zoomed-out perspective on innovation, publications are a coarse-grained record of individuals’ insights. Explaining the sudden insights of innovators, from scientists to sculptors, requires attending to the local, distributed systems of their intellectual activity.
General Social Agents
Benjamin Manning & John Horton
MIT Working Paper, September 2025
Abstract:
Useful social science theories predict behavior across settings. However, applying a theory to make predictions in new settings is challenging: rarely can it be done without ad hoc modifications to account for setting-specific factors. We argue that AI agents put in simulations of those novel settings offer an alternative for applying theory, requiring minimal or no modifications. We present an approach for building such “general” agents that use theory-grounded natural language instructions, existing empirical data, and knowledge acquired by the underlying AI during training. To demonstrate the approach in settings where no data from that data-generating process exists -- as is often the case in applied prediction problems -- we design a highly heterogeneous population of 883,320 novel games. AI agents are constructed using human data from a small set of conceptually related, but structurally distinct “seed” games. In preregistered experiments, on average, agents predict human play better than (i) game-theoretic equilibria and (ii) out-of-the-box agents in a random sample of 1,500 games from the population. For a small set of separate novel games, these simulations predict responses from a new sample of human subjects better even than the most plausibly relevant published human data.
Beggars as Rational Choosers
Peter Leeson, August Hardy & Paola Suarez
Southern Economic Journal, forthcoming
Abstract:
American municipalities increasingly regulate panhandling. That regulation is controversial. The determinants of panhandling activeness are unknown, and it is doubted whether panhandling activity responds rationally to incentives. To shed light on these issues, we collect data on hundreds of panhandlers and the passersby they solicit at Metrorail stations in Washington, DC. Consistent with a simple model of profit-maximizing panhandling, we find that panhandlers solicit more actively when they compete, when they have more human capital, and when passersby are more numerous and responsive to solicitation. Beggars are choosers, and they appear to be rational ones.