RESUMEN
Large-scale social networks are thought to contribute to polarization by amplifying people's biases. However, the complexity of these technologies makes it difficult to identify the mechanisms responsible and evaluate mitigation strategies. Here we show under controlled laboratory conditions that transmission through social networks amplifies motivational biases on a simple artificial decision-making task. Participants in a large behavioural experiment showed increased rates of biased decision-making when part of a social network relative to asocial participants in 40 independently evolving populations. Drawing on ideas from Bayesian statistics, we identify a simple adjustment to content-selection algorithms that is predicted to mitigate bias amplification by generating samples of perspectives from within an individual's network that are more representative of the wider population. In two large experiments, this strategy was effective at reducing bias amplification while maintaining the benefits of information sharing. Simulations show that this algorithm can also be effective in more complex networks.
Asunto(s)
Algoritmos , Red Social , Humanos , Teorema de Bayes , Sesgo , MotivaciónRESUMEN
Many of the computational problems people face are difficult to solve under the limited time and cognitive resources available to them. Overcoming these limitations through social interaction is one of the most distinctive features of human intelligence. In this paper, we show that information accumulation in multigenerational social networks can be produced by a form of distributed Bayesian inference that allows individuals to benefit from the experience of previous generations while expending little cognitive effort. In doing so, we provide a criterion for assessing the rationality of a population that extends traditional analyses of the rationality of individuals. We tested the predictions of this analysis in two highly controlled behavioral experiments where the social transmission structure closely matched the assumptions of our model. Participants made decisions on simple categorization tasks that relied on and contributed to accumulated knowledge. Success required these microsocieties to accumulate information distributed across people and time. Our findings illustrate how in certain settings, distributed computation at the group level can pool information and resources, allowing limited individuals to perform effectively on complex tasks.