This paper explores the phenomenon of information cascades in a laboratory setting, where individuals with private information make public predictions. The authors conduct an experiment where subjects draw from urns with different probabilities of containing either an "a" or "b" signal, and they are paid for correctly guessing which urn was used. The study finds that initial conformity can lead to rational cascades, where subsequent decision-makers follow the established pattern despite their private information. However, the presence of errors and biases can disrupt these cascades, leading to reverse cascades where incorrect initial signals cause a chain of incorrect decisions. The paper also discusses the role of status quo bias and the tendency to rely on simple counting heuristics rather than Bayesian reasoning. Overall, the results suggest that information cascades are robust but can be influenced by various cognitive and behavioral factors.This paper explores the phenomenon of information cascades in a laboratory setting, where individuals with private information make public predictions. The authors conduct an experiment where subjects draw from urns with different probabilities of containing either an "a" or "b" signal, and they are paid for correctly guessing which urn was used. The study finds that initial conformity can lead to rational cascades, where subsequent decision-makers follow the established pattern despite their private information. However, the presence of errors and biases can disrupt these cascades, leading to reverse cascades where incorrect initial signals cause a chain of incorrect decisions. The paper also discusses the role of status quo bias and the tendency to rely on simple counting heuristics rather than Bayesian reasoning. Overall, the results suggest that information cascades are robust but can be influenced by various cognitive and behavioral factors.