I think living state, score and x,y position is all you would need to train a player. Is there any advantage to consuming the state of nearby cells if the game

I think living state, score and x,y position is all you would need to train a player. Is there any advantage to consuming the state of nearby cells if the game is entirely deterministic?