The right decision today may be the wrong decision tomorrow. We live in a world in which expectations, contingencies, and goals continually evolve and change. Thus, decisions do not occur in isolation but rather are tightly embedded in these streams of temporal dependencies. Accordingly, even relatively straightforward visual decisions must take into account not just the immediate sensory input but also past experiences and future goals and expectations. Here, we evaluate recent progress in understanding how the brain implements these dependencies. We show that visual decision-making relies on mechanisms of evidence accumulation and commitment that have been studied extensively under relatively static, isolated conditions but in general can operate much more flexibly. A deeper understanding of these mechanisms will require identifying the principles that govern this flexibility, which must operate across different timescales to produce effective decisions in uncertain and dynamic environments.
Keywords: Bayesian decision theory; perception; priors; psychophysics; sequential analysis; signal detection theory.