We introduce a model of managerial problem formulation which proposes that a key decision criterion for choosing among alternative managerial problem statements is the computational complexity (K-complexity) of the required solution algorithm of alternative problem statements. We distill many typical managerial problems to their canonical algorithmic structures and group them into classes on the basis of their K-complexity. P-type problems (canonically easy ones) require solution algorithms of complexity that is at most a polynomial function of the number of the variables of the problem, NP-type problems (canonically hard ones) require solution algorithms of complexity that is a higher-than-any-polynomial function of the number of problem variables. We show that a model in which managers prefer to solve P-hard problems over solving NP-hard problems explains many observed patterns of managerial behavior and cognition. We consider within-complexity-class choices among alternative problem statements and introduce the notion of computational landscapes to examine the marginal costs and benefits of additional calculative thinking in situations where competitors are either more or less inclined to engage in computationally deep reasoning tasks.