Prolonging network lifetime by salvaging the energy of low battery capacity nodes is a prime concern in Wireless Sensor Networks. Energy efficient routing protocols try to improve the node lifetime by restricting their participation in the routing process. This leads to selfish node behavior causing disruption of inherent network cooperation. This paper proposes an elegant routing mechanism based on direct reciprocity principle, named energy efficient cooperative ad hoc on-demand distance vector (EECoAODV). Proposed protocol correctly differentiates between inherently selfish nodes that use the energy of other nodes to relay their packets, but refuse to reciprocate; and the energy critical nodes that have turned non-cooperative for their own survival. Selfish nodes are punished and eliminated from the routing process thus prolonging the battery capacity of energy critical nodes to improve the overall network performance. EECoAODV is implemented in Qualnet simulator and its performance is compared with conventional AODV and reinforcement based state-action-reward-action (SARSA) routing mechanism. Results show that EECoAODV improves the lifetime of energy critical nodes and thus delivers improved packet delivery ratio than SARSA-AODV and conventional AODV.