Object-specific　selective　attention　can　be　achieved　either　by　simultaneously　splitting　attention　to　multiple　objects,　or　by　sequentially　shifting　spatial　attention　among　objects.　A　growing　body　of　research　show　that　object-specific　selective　attention　can　be　implemented　using　the　second　way　and　that　the　sequential　movement　of　attention　exhibits　specific　rhythmicity.　However,　the　neurocomputing　mechanisms　underlying　this　phenomenon　are　still　not　fully　understood.　To　clarify　this　issue,　we　conducted　magnetoencephalography　experiments　on　healthy　participants　and　subsequently　proposed　a　computational　framework　based　on　time-series　decomposition　and　rhythmic　analysis　to　delve　into　the　neural　mechanisms　of　object-specific　selective　attention.　Our　investigation　reveals　that　the　four　single-object　attention　states　are　decodable　on　the　level　of　magnetoencephalography　(MEG)　sensor　signals.　Furthermore,　these　states　manifest　dynamically　and　rhythmically　during　object-specific　selective　attention.　These　findings　suggest　that　the　attentional　rhythm　exhibited　by　neural　activity　during　object-specific　selective　attention　is　fundamentally　characterized　by　a　set　of　basic　attentional　units.　This　research　provides　valuable　information　for　future　investigations　into　the　brain　model　of　object-specific　selective　attention.　©　2025