This thesis is about multi-period problems in which the decision-maker or players cannot see far enough ahead to solve the problem completely. The thesis considers why it might be that players reason forwards at all, let alone reasoning forwards only finitely far. It shows, using finite automata, that there is a class of problems for which forwards reasoning is more efficient than backwards reasoning. It goes on to use these finite automata to solve for an optimal foresight length. It then discusses solution concepts, and applies its preferred solution concept to two problems - one macro problem involving a central banker, and one micro problem concerning the decision whether to smoke.