Playing a symmetric bi-matrix game is usually physically implemented by sharing pairs of 'objects' between two players. A new setting is proposed that explicitly shows effects of quantum correlations between the pairs on the structure of payoff relations and the 'solutions' of the game. The setting allows a re-expression of the game such that the players play the classical game when their moves are performed on pairs of objects having correlations that satisfy the Bell's inequalities. If players receive pairs having quantum correlations the resulting game cannot be considered another classical symmetric bi-matrix game. Also the Nash equilibria of the game are found to be decided by the nature of the correlations.
I. INTRODUCTIONPlaying a game requires resources for its physical implementation. For example, to play a bi-matrix game the resources may consist of pairs of two-valued 'objects', like coins, distributed between the players. The players perform their moves on the objects and later a referee decides payoffs after observing them. Game theory usually links players' actions directly to their payoffs, without a reference to the nature of the objects on which the players have made their moves. Analysis of quantum games [1] suggests radically different 'solutions' can emerge when the same game is physically implemented on distributed objects which are quantum mechanically correlated.
Much of recent work on quantum games [1-4] uses a particular quantization scheme[1] developed for a bi-matrix game where two players, on receiving an entangled two-qubit state, play their moves by local and unitary actions on the state. After disentanglement, a measurement of the state rewards the players their payoffs. The payoffs become classical when the moves are performed on a product state. For example, in Prisoners' Dilemma new and more beneficial equilibrium emerges in its quantum form [1] when the allowed moves