The allocation of time across competing tasks di§ers from many other resource allocation problems
in one important respect: time is irreversible and cannot be reallocated once it has elapsed. If
a decision maker has both the cognitive sophistication and the information necessary to compute
optimal allocations, this di§erence is inconsequential. When information about the underlying
parameters of the problem is incomplete, however, the irreversibility of time precludes the use of
certain strategies, such as allocation by trial and error. In this case, one must rely on heuristics
based on information that becomes available during the process of allocation itself. Under what
conditions might one expect such heuristics to result in optimal allocations? This is the main
question addressed in this paper.