This paper investigates the interplay and decision-making consequences of biases in a manager’s initial beliefs and belief-updating behavior, with a focus on asymmetric biases in both domains. A manager’s bias in initial beliefs manifests as either overestimation or underestimation of future out- comes. A manager’s belief-updating can be biased by predominantly learning from feedback that is better than expected while discounting feedback that is lower or by learning from feedback that is worse than expected and discounting such that exceeds expectations. Using a multi-arm bandit model, we examine the spectrum of biases as types of pessimism and optimism, which can exist in any initial belief and belief-updating combination. The distinction and interaction between biases in beliefs and belief-updating demonstrate that neither is inherently detrimental and can indeed lead to high-performing decision-making in appropriate circumstances. Performance in optimistic belief-updating depends on the manager’s commitment to exploration, while the efficacy of pessimistic belief-updating is enhanced by initially optimistic beliefs.
