Cognition capability has been seen by researchers as the way forward for the design of next generation of Mobile Ad Hoc Networks (MANETs). The reason why a cognitive paradigm would be more suited to a MANET is because MANETs are highly dynamic networks. The topology may change very frequently during the operation of a MANET. Traffic patterns in MANETs can vary from time to time depending on the need of the users. The size of a MANET and node density is also very dynamic and may change without any predictable pattern. In a MANET environment, most of these parameters may change very rapidly and keeping track of them manually would be very difficult. Previous studies have shown that the performance of a certain routing approach in MANETs is dependent on the size of the network and node density. The choice of whether to use a reactive or proactive routing approach comes down to the network size parameter. Static or offline approaches to fine tune a MANET to achieve certain performance goals is hence not very productive as a lot of these parameters keep changing during the course of operation of MANETs. Similarly, the performance of MANETs would improve greatly if the MAC layer entity could operate in a more flexible manner. In this paper we propose a cognitive MANET design that will ensure that all these dynamic parameters are automatically monitored and decisions are based on the current status of these parameters.