We consider the problem of communication over a multiple access channel (MAC) with noiseless feedback. A single-letter characterization of the capacity of this channel is not currently known. Several achievable schemes exist in the literature that provide achievable region expressions involving severeal auxiliary variables. On the other hand, a series of recent results in point-to-point communications with noiseless feedback study the information theoretic problem of finding channel capacity in a stochastic control framework. This approach has not yet been fruitful in the study of multi-user communication scenaria. In this paper we formulate the MAC with feedback capacity problem as a stochastic control problem. This interpretation provides an understanding of the role of auxiliary random variables. In addition, through this interpretation a useful outer bound is derived that is very close to known achievable schemes. For the special case when the Cover and Leung region is optimal this outer bound is tight.