The reaction function stems from the realization that an oligopolist is aware of the presence and activities of other firms in the market and the realization that, over time, changes in the behavior of a firm are likely to induce changes in the behavior of its rivals. At first it may seem impossible to discuss the behavior of a small number of interacting decision makers without talking in terms such as these: If A does a, then B will do b; but if B does b, then A will do a′; but if A does a′ then B will do b′ and so forth. Such discussion is carried out within the framework of single-period models, where often it is logically inconsistent or is based on economically untenable assumptions, by writers going back to Cournot. The reason for persisting in this vein is that what is most fascinating about oligopoly is the intertemporal interaction of the firms in a market. What we most wish to understand is how the firms behave in their own best interests, taking into account rival actions, and it seems impossible to do this unless each firm's plan of action (strategy) prescribes what to do in each time period as a function of the recently observed choices of rival firms.

If this has often been done in the past, and to some extent still is done, using models that formally are only single-period models, it is because many writers have found multiperiod models to be intractable.