Decision making is an ubiquitous and important component of behaviour. For decades, the dominant theoretical model of decision making is that of “evidence accumulation to bound”. This model entails that in order to commit to a choice, people integrate evidence for that choice over time, until they reach a critical boundary value, which triggers a response. Computational models of this mechanism have been extremely successful in explaining a wide array of behavioural as well as neural phenomena related to decision-making. Recently, a debate has sparked around the nature of the critical boundary value that triggers a choice. Recent evident suggests that the boundary may be time-varying, meaning that while time progresses, participants are more likely to commit to a choice on the basis of less information. However, at the same time models that assume a time-invariant (static) bound remain highly successful in explaining data. In my talk I will go into this apparent discrepancy, and show that (1) a static bound can be a very good approximation of a time-varying bound when participants are asked to optimize their rate of reward; (2) Participants show evidence of static as well as time-varying boundaries, depending on the experimental manipulation; and (3) the neural mechanisms that are involved in static and time-varying boundaries are shared, suggesting that participants have control over the decision strategy, and engage a different boundary setting depending on the experimental context.