Authors: Jonas Stein, Shannon Cruz, Davide Grossi, Martina Testori
Organizations: Faculty of Behavioural and Social Sciences, University of Groningen, The Netherlands. · Department of Communication Arts and Sciences, The Pennsylvania State University, USA · Bernoulli Institute for Mathematics, Computer Science and Artificial Intelligence, University of Groningen; Amsterdam Center for Law & Economics, University of Amsterdam; Institute for Logic, Language and Computation, University of Amsterdam, The Netherlands. · Greenwich Business School, University of Greenwich, Old Royal Naval College, UK
A core tenet underpinning the conception of contemporary information networks, such as social media platforms, is that users should not be constrained in the amount of information they can freely and willingly exchange with one another about a given topic. By means of a computational agent-based model, we show how even in groups of truth-seeking and cooperative agents with perfect information-processing abilities, unconstrained information exchange may lead to detrimental effects on the correctness of the group's beliefs. If unconstrained information exchange can be detrimental even among such idealized agents, it is prudent to assume it can also be so in practice. We therefore argue that constraints on information flow should be carefully considered in the design of communication networks with substantial societal impact, such as social media platforms.
Figures & tables
Figure 1: Each agent is represented by a colored figure, with the color indicating their belief: Green-leaning agents are more confident that the true state is A , while red-leaning agents are more confident in the false state B . Step 1: Initial state ( t = 0). Each agent is endowed with a private piece of evidence about the state of the world. Each piece of evidence (green lightbulb for A , red lightbulb for B ) comes with a known quality — the probability that the signal is accurate (full/empty lightbulb for high/low quality). Agents use Bayes’ rule to form an initial belief based on this single observation. Some agents start with a correct belief (favoring A ), others with an incorrect one (favoring B ) due to randomness in their initial evidence. Step 2: Partner selection. At each time step, agents are stochastically matched with another agent to exchange information. The likelihood of being paired increases with belief similarity, a dynamic controlled by homophily: When homophily is high, agents are more likely to interact with others who hold beliefs similar to their own. Step 3: Belief exchange. Agents exchange up to k pieces of evidence, determined by their communication capacity. Each agent selects their best available evidence — i.e., the most accurate signals — and shares them in proportion to their current belief. For example, an agent strongly favoring A will mostly share high-quality signals supporting A , but may also share some evidence supporting B . Note: Exchanges occur in one direction, meaning that one agent might not share information with the same agent they received information from. Step 4: Belief update. After the exchange, each agent updates their belief by applying Bayes’ rule to the new evidence received. Over time, these updates can lead agents to converge toward the true state A , though this depends on both who they interact with (influenced by homophily) and how much information they can exchange (communication capacity).
Figure 2: Belief evolution over interactions by homophily h and communication capacity k . Panel i & iii : Low communication capacity results in moderate belief diversity and epistemic performance, regardless of the level of homophily. Panel ii: High communication capacity and high homophily result in belief polarization and low epistemic performance. Panel iv: Epistemic performance is maximal at high capacity and no homophily. Green (red) lines represent agents with an initial belief in favor of A ( B ).
Figure 3: Overview of a) agents’ A and B epistemic gain ( ρi ; first column); b) groups’ A and B epistemic gain ( WA,WB ; middle column); c) population epistemic gain ( W ; top panel, last column); d) inequality across the population in the epistemic gain ( G ; bottom panel, last column), as a function of homophily h and communication capacity k .
Social media platforms increasingly use AI-based credibility indicators to help users judge misinformation. Unlike individual human-AI decision-making, these indicators are embedded in information spread: users see both an AI prediction and earlier judgments shaped by the same AI, and their own judgments may then enter the public history. Yet how to analytically characterize this process remains under-explored. We therefore introduce a social-learning lens for this setting by extending the classical Bayesian cascade model with the AI indicator as a shared public signal. The resulting Gateway condition compares the evidence from the AI prediction with users' private impressions. Through this view, we show that AI changes what public history means. Crowd agreement may reflect accumulated independent human evidence, or repeated dependence on the same AI prediction. This creates a preservation-correction trade-off: stronger reliance on AI can preserve correct predictions, but can also lock in incorrect ones by blocking corrective private impressions. We calibrate the model using human-subject data on news veracity judgments. Although the AI outperforms human users, the average user weights it below her own impression but above several peer judgments, while individual users vary from discounting the AI to relying on it enough to cascade. Simulations show that over-reliance on a weak AI is especially harmful, and that diversifying AI signals across users can better keep the crowd informative. We conclude with implications for understanding human-AI interaction in information spread and designing misinformation interventions.
Zhuoran Lu, Weilong Wang, Yangyang Yu +4
Purdue University West Lafayette, Indiana, USA · St John’s University New York, New York, USA · Stevens Institute of Technology Hoboken, New Jersey, USA +2
Organizations often pool dispersed information into one ranking and then allow many agents to act on that shared view. In a discovery problem, this can improve beliefs while reducing coverage. We develop an exactly solvable benchmark with sixteen boxes, one target, eight searchers, and noisy private clues. Pooling raises the accuracy of the best single recommendation from 0.20 to 0.3835, but repeating that recommendation lowers group discovery from 0.8322 under decentralized clue-following to 0.3835. A coordinated eight-action portfolio using the same pooled reports reaches 0.8594, and seven coordinated actions recover the decentralized benchmark. The paradox is a protocol failure, not an information failure: a one-answer rule compresses a portfolio of available actions into one repeated choice. We then replace the planner with self-interested searchers who split a prize. The equal-split game is a potential game. Its anonymous symmetric equilibrium obeys a water-filling rule. In the canonical instance it achieves 0.5991: strictly above consensus, but below both private search and the planner. The exact mixed price of anarchy is 2 - 1/N. A sole-rescue reward, which pays only an agent who covers the target alone, makes every pure Nash equilibrium first-best. Finally, a latent common-cue model shows how correlated reports collapse effective discovery channels. The centralized planner gain rises strictly with copying, and in the canonical environment the symmetric market overtakes decentralized report-following at copying probability c = 0.788462. In a proportional large-market limit the five-protocol ordering survives exactly: consensus discovery vanishes while blind, market, private, and portfolio search converge to 0.500, 0.547, 0.847, and 0.874. The contribution is a compact benchmark that separates information, allocation, incentives, and dependence into exact, reusable quantities.
In this (work in progress) paper, we present Bounded Pragmatic Listener (or BPL), a cognitively grounded Bayesian framework for modelling susceptibility to information disorder. BPL extends Rational Speech Act theory with three cognitively motivated bounds derived from the bounded rationality literature with a) a recursion depth bound (that emphasises working memory limits);b) a prior compression parameter (which is oriented at capturing information bottleneck); and c) an availability sample size (that operationalises importance sampling with saliency-weighted proposals). This allows us to test predictions about misinformation susceptibility, annotator disagreement, and the differential vulnerability to mis-, dis-, and mal-information as defined in the Information Disorder framework. We validate BPL on the LIAR and MultiFC benchmarks showcasing competitive veracity classification and experimental support for the depth-mismatch paradox.
Pranava Madhyastha
Dept. of Computer Science, City, University of London · The Alan Turing Institute