In　the　traditional　setup　of　evolutionary　game　theory　such　as　the　Public　Goods　Game　(PGG),　all　players　are　involved　in　every　available　group　and　the　game　relationships　between　individuals　are　undirected　and　symmetric.　Inspired　by　the　exit　mechanism　in　the　realistic　PGG,　i.e.,　that　players　would　unilaterally　choose　to　quit　playing　if　the　payoffs　do　not　meet　their　expectations,　we　introduce　directionality　and　asymmetry　into　the　game　relationships　between　players,　and　propose　both　active　and　dormant　mechanisms　of　individual　game　relationships.　In　this　new　scenario,　each　player　would　choose　to　keep　unilaterally　dormant　when　his　or　her　payoff　is　less　than　the　expected　value,　and　become　reactivated　when　certain　conditions　are　satisfied.　Our　results　reveal　an　optimal　region　of　the　expected　payoff　and　the　recovery　rate,　corresponding　to　the　highest　frequency　of　cooperation　in　the　PGG.　Spatial　distributions　of　active　and　dormant　groups　are　obtained　to　understand　the　evolution　of　cooperation.